Word-based nvtext::minhash function

[davidwendt]

Description

Experimental implementation for #15055
The input is a lists column of strings where each string in each row is expected as a word to be hashed. The minimum hash for that row is returned in a lists column where each row contains a minhash per input hash seed.
Here the caller is expected to produce the words to be hashed.

std::unique_ptr<cudf::column> word_minhash(
  cudf::lists_column_view const& input,
  cudf::device_span<uint32_t const> seeds,
  rmm::cuda_stream_view stream,
  rmm::device_async_resource_ref mr);

Checklist

• I am familiar with the Contributing Guidelines.
• New or existing tests cover these changes.
• The documentation is up to date with these changes.

[davidwendt]

/merge