Count distinct floats

[pjmore]

Which issue does this PR close?

Closes#199 .

What changes are included in this PR?

Modified try_from_array method on ScalarValue and added tests to physical_plan/distinct_expressions.rs

The tests cover all of the floating point weirdness that I could think of, namely NaN, the infinities, and subnormal numbers, but if I missed any I'm happy to add tests for them.

Are there any user-facing changes?

No

[Dandandan]

Looks good, thanks @pjmore !

[codecov-commenter]

Codecov Report

Merging #252 (774d7b4) into master (e271e4d) will increase coverage by 0.00%.
The diff coverage is 70.00%.

@@ Coverage Diff @@## master #252 +/- ##
=======================================
Coverage 76.80% 76.81% =======================================
Files 133 133 Lines 23284 23294 +10 =======================================
+ Hits 17884 17894 +10 
Misses 5400 5400 

Impacted Files	Coverage Δ
datafusion/src/scalar.rs	54.36% <0.00%> (+0.22%)	⬆️
...tafusion/src/physical_plan/distinct_expressions.rs	90.80% <87.50%> (-0.11%)	⬇️
datafusion/src/physical_plan/group_scalar.rs	58.82% <0.00%> (+1.17%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e271e4d...774d7b4. Read the comment docs.

[alamb]

I tried it out locally and it works great. Thank you @pjmore !

alamb@ip-10-0-0-124:~/Software/arrow-datafusion$ echo "foo,1.23" > /tmp/foo.csv
alamb@ip-10-0-0-124:~/Software/arrow-datafusion$ cargo run -p datafusion-cli
Finished dev [unoptimized + debuginfo] target(s) in 0.13s
Running `target/debug/datafusion-cli`
> CREATE EXTERNAL TABLE t (a varchar, b float) STORED AS CSV LOCATION '/tmp/foo.csv';
0 rows in set. Query took 0 seconds.
> select count(distinct a) from t;
+-------------------+
| COUNT(DISTINCT a) |
+-------------------+
| 1 |
+-------------------+
1 rows in set. Query took 0 seconds.
> select count(distinct b) from t;
+-------------------+
| COUNT(DISTINCT b) |
+-------------------+
| 1 |
+-------------------+
1 rows in set. Query took 0 seconds.

[Dandandan]

Thanks again @pjmore !

[alamb]

🎉