Allow flag to alias all projected substrait expressions with a UUID

[xanderbailey]

Which issue does this PR close?

• ClosesExpressions from left and right of a join can fail planning during substrait converstion if their name is the same #19330

Rationale for this change

As previously discussed in #17299. This continues to be a bit of a thorn for expressions that are the same on the left and the right side of the of a join leading to ambiguous references.

Downside here is that it makes the plan less readable which is why I thought it would be better if this was a config that users can opt into to make things more stable. Having said that in my experience the plans created via substrait are super hard to read anyway. If you have a deeply nested CAST statement for example then the name that appears in the plan is unintelligible anyway.

What changes are included in this PR?

Adds substrait_alias_all_expressions config which applies a UUID alias to all expressions during substrait conversion.

Are these changes tested?

Yes

Are there any user-facing changes?

[xanderbailey]

Happy to take ideas on the best way to write this test but I wanted to show that the UUIDs actually light up correctly.

[xanderbailey]

@LiaCastaneda thought you might have some ideas here

[LiaCastaneda]

👋 Sorry for the late reply,
I think this fixes all the issues once for all, however I'm wondering if maybe it would be worth it to make this fix in the name tracker instead? as suggested in #17508. From my understanding what would have to be done is basically track two fields which are the schema name and the qualified name.
cc @dd-annarose since you were looking into this.

[xanderbailey]

I'm happy if there's a better route forward we're just looking for a quick fix without having to fork the whole substrait conversation.

[xanderbailey]

Had a go at the fix here #19856