Adds explode API

[isVoid]

Closes #2975

This PR introduces explode API, which flattens list columns and turns list elements into rows. Example:

>>> s = cudf.Series([[1, 2, 3], [], None, [4, 5]])
>>> s
0    [1, 2, 3]
1           []
2         None
3       [4, 5]
dtype: list
>>> s.explode()
0       1
0       2
0       3
1    <NA>
2    <NA>
3       4
3       5
dtype: int64

Supersedes #7538

[isVoid]

rerun tests

[codecov]

Codecov Report

Merging #7607 (e5e83b1) into branch-0.19 (7871e7a) will increase coverage by 0.62%.
The diff coverage is 93.33%.

❗ Current head e5e83b1 differs from pull request most recent head 6ce751a. Consider uploading reports for the commit 6ce751a to get more accurate results

@@               Coverage Diff               @@
##           branch-0.19    #7607      +/-   ##
===============================================
+ Coverage        81.86%   82.49%   +0.62%     
===============================================
  Files              101      101              
  Lines            16884    17397     +513     
===============================================
+ Hits             13822    14351     +529     
+ Misses            3062     3046      -16     

Impacted Files	Coverage Δ
python/cudf/cudf/core/index.py	93.45% <ø> (+0.59%)	⬆️
python/cudf/cudf/core/series.py	91.69% <ø> (+0.90%)	⬆️
python/cudf/cudf/core/tools/datetimes.py	84.53% <ø> (+0.08%)	⬆️
python/cudf/cudf/utils/cudautils.py	52.94% <ø> (+2.55%)	⬆️
python/cudf/cudf/utils/dtypes.py	89.88% <ø> (+0.37%)	⬆️
python/dask_cudf/dask_cudf/io/orc.py	91.04% <ø> (+0.13%)	⬆️
python/cudf/cudf/core/column/numerical.py	94.83% <87.50%> (-0.20%)	⬇️
python/cudf/cudf/core/column/column.py	87.86% <90.00%> (+0.10%)	⬆️
python/cudf/cudf/core/column/decimal.py	92.75% <90.32%> (-2.12%)	⬇️
python/cudf/cudf/core/frame.py	89.23% <91.83%> (+0.21%)	⬆️
... and 58 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 0b766c5...6ce751a. Read the comment docs.

[kkraus14]

Pandas also allows calling explode against non-list columns which seems to return the Series / DataFrame by copy. We may want to handle the same for compatibility.

[kkraus14]

Do we not have a helper function that handles all of the "gotchas" in this?

[isVoid]

Checked with @shwina, and can't think of any.. ._data is constructed inside from_unique_ptr, which has no information about multiindex and _level_names.

[shwina]

I agree it's not ideal. One alternative I've considered in the past is a custom ColumnMeta type containing information about column names:

class ColumnMeta:
    names: Tuple[Any]
    multiindex: bool
    level_names: Tuple[Any]

And we pass objects of that type as arguments to ._from_unique_ptr:

cdef from_unique_ptr(unique_ptr[table] c_tbl, ColumnMeta data_meta, ColumnMeta index_meta=None):
    pass

Maybe a bit out of scope for this PR though.

[kkraus14]

Definitely out of scope for this PR, but I like it a lot.

[isVoid]

Is ColumnMeta owned and maintained by some class? Or is it a message interface that gets created on the fly and passes around classes?

[shwina]

Good question -- I think we should have a separate discussion about this outside the context of this PR. Happy to sync offline and then raise an issue?

[kkraus14]

One minor doc issue, otherwise LGTM!

[galipremsagar]

Minor doc fixes needed, else LGTM

[kkraus14]

@gpucibot merge