[REVIEW] Add filter_tokens nvtext API

[davidwendt]

Closes #5521

Add function to filter tokens from strings based on the length of each token as described in #5521

std::unique_ptr<cudf::column> filter_tokens(
  cudf::strings_column_view const& strings,
  cudf::size_type min_token_length,
  cudf::string_scalar const& replacement = cudf::string_scalar{""},
  cudf::string_scalar const& delimiter   = cudf::string_scalar{""},
  rmm::mr::device_memory_resource* mr    = rmm::mr::get_default_resource());

This PR also includes the Python strings API and the required Cython interfaces.

    def filter_tokens(
        self, min_token_length, replacement=None, delimiter=None, **kwargs
    ):

[codecov]

Codecov Report

Merging #5658 into branch-0.15 will decrease coverage by 0.27%.
The diff coverage is 100.00%.

@@               Coverage Diff               @@
##           branch-0.15    #5658      +/-   ##
===============================================
- Coverage        86.21%   85.94%   -0.28%     
===============================================
  Files               72       72              
  Lines            12727    12439     -288     
===============================================
- Hits             10973    10691     -282     
+ Misses            1754     1748       -6     

Impacted Files	Coverage Δ
python/cudf/cudf/core/column/string.py	86.61% <100.00%> (+0.03%)	⬆️
python/cudf/cudf/io/hdf.py	50.00% <0.00%> (-7.15%)	⬇️
python/cudf/cudf/utils/cudautils.py	46.80% <0.00%> (-3.53%)	⬇️
python/cudf/cudf/core/abc.py	88.09% <0.00%> (-3.21%)	⬇️
python/cudf/cudf/comm/gpuarrow.py	87.67% <0.00%> (-1.62%)	⬇️
python/cudf/cudf/core/column/datetime.py	84.41% <0.00%> (-1.04%)	⬇️
python/cudf/cudf/core/multiindex.py	80.11% <0.00%> (-0.77%)	⬇️
python/cudf/cudf/core/buffer.py	82.00% <0.00%> (-0.70%)	⬇️
python/cudf/cudf/core/index.py	91.17% <0.00%> (-0.59%)	⬇️
python/cudf/cudf/utils/utils.py	84.40% <0.00%> (-0.41%)	⬇️
... and 26 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 1f4209c...39a4409. Read the comment docs.

[kkraus14]

Python / Cython looks good

[trevorsm7]

Looks great. Good use of 🔥 in the unit tests 😄. Just a minor suggestion to try empty_like

[vuule]

A few minor questions/suggestions, the PR looks great overall!