Allow casting Timestamp arrays into String

[sum12]

Which issue does this PR close?

Closes #587

Rationale for this change

What changes are included in this PR?

there are two commits included in the PR
the first commit shows how it should be possible to implement. But the commit fails to compile with

error[E0599]: the method `value_as_datetime` exists for reference `&array_primitive::PrimitiveArray<T>`, but its trait bounds were not satisfied
    --> arrow/src/compute/kernels/cast.rs:1044:48
     |
1044 |     (0..array.len()).for_each(|ix| match array.value_as_datetime(ix) {
     |                                                ^^^^^^^^^^^^^^^^^ method cannot be called on `&array_primitive::PrimitiveArray<T>` due to unsatisfied trait bounds
     |
     = note: the following trait bounds were not satisfied:
             `T: numeric::ArrowNumericType`

error[E0599]: the method `value_as_datetime` exists for reference `&array_primitive::PrimitiveArray<T>`, but its trait bounds were not satisfied
    --> arrow/src/compute/kernels/cast.rs:1044:48
     |
1044 |     (0..array.len()).for_each(|ix| match array.value_as_datetime(ix) {
     |                                                ^^^^^^^^^^^^^^^^^ method cannot be called on `&array_primitive::PrimitiveArray<T>` due to unsatisfied trait bounds
     |
     = note: the following trait bounds were not satisfied:
             `T: numeric::ArrowNumericType`

error: aborting due to previous error

and compiles fine with second commit. But it feels not correct to use the second commit as the ArrowNumericType and Timestamp* types are not really releated. (Or I do not understand them correctly)

Are there any user-facing changes?

yes (additional ways to cast array) (where do I document this ?)
no breaking changes

[alamb]

Thanks @sum12 -- I plan to check this out tomorrow

[alamb]

This is looking good @sum12 . I left some style comments -- but the only thing this PR really needs to be good to go is some basic tests. Perhaps you can follow the style of the existing tests?

[nevi-me]

happy that we keep nulls now

[nevi-me]

LGTM @alamb I'll leave it to you to have a final look + merge

[codecov-commenter]

Codecov Report

Merging #664 (d9de2de) into master (b38a4b6) will decrease coverage by 0.04%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #664      +/-   ##
==========================================
- Coverage   82.50%   82.45%   -0.05%     
==========================================
  Files         168      168              
  Lines       47237    47379     +142     
==========================================
+ Hits        38971    39066      +95     
- Misses       8266     8313      +47     

Impacted Files	Coverage Δ
arrow/src/compute/kernels/cast.rs	94.63% <100.00%> (+0.09%)	⬆️
arrow/src/array/transform/utils.rs	95.00% <0.00%> (-5.00%)	⬇️
arrow/src/array/equal_json.rs	85.21% <0.00%> (-3.48%)	⬇️
arrow/src/tensor.rs	85.00% <0.00%> (-2.50%)	⬇️
parquet/src/column/page.rs	97.36% <0.00%> (-1.32%)	⬇️
arrow/src/array/equal/utils.rs	74.00% <0.00%> (-1.00%)	⬇️
parquet/src/record/api.rs	91.60% <0.00%> (-0.88%)	⬇️
parquet/src/file/statistics.rs	93.80% <0.00%> (-0.83%)	⬇️
arrow/src/csv/writer.rs	82.41% <0.00%> (-0.74%)	⬇️
arrow/src/array/array_map.rs	81.50% <0.00%> (-0.69%)	⬇️
... and 22 more

---

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 b38a4b6...d9de2de. Read the comment docs.

[alamb]

This looks great ! Thank you again for your contribution @sum12 -- I am sorry for the delay in reviewing.