split a string basic type away from symbolBase

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201

[perrydv]

Support for strings is still rudimentary. In issue #201 we saw that even returning a string triggered a warning because the type checking for return objects was based on the basic types of double, integer, and logical. However the string type also hijacked the same symbol type, resulting in an error in the Return error checking. Strings are a somewhat hybrid case because there is some commonality, especially if we can use Eigen Tensors to store and access non-scalars for strings and possible other kinds of elements. So this PR takes the approach of inserting a layer into the symbol class hierarchy. The new layer serves as a base class for any core type that might be scalar or not. Then symbolBasic (for double, integer, and logical, and also AD) derives from that, and a new symbolBasicString also derives from that. It meant that all code checking on or working with the symbolBasic type was not broken.

In addition I separated void into its own symbol class rather than also hijacking a special case of symbolBasic for this.

Tests passed, so this looks good to go.

Fixes#201