feat: Hugging Face path expansion

[nameexhaustion]

This should allow us to scan Parquet, CSV and IPC from hf://

r = pl.scan_parquet(
    "hf://datasets/roneneldan/TinyStories/data/train-00000-of-00004-2d5a1467fff1081b.parquet"
).head(1)

print(r.collect())

r = pl.scan_parquet(
    "hf://datasets/roneneldan/TinyStories@~parquet/default/train/*.parquet"
).head(1)

print(r.collect())

r = pl.scan_csv("hf://datasets/fka/awesome-chatgpt-prompts/*.csv")

print(r.collect())

ref #17625

[codecov]

Codecov Report

Attention: Patch coverage is 53.28947% with 142 lines in your changes missing coverage. Please review.

Project coverage is 80.71%. Comparing base (08b6f1d) to head (ad2fb1a).
Report is 2 commits behind head on main.

Files	Patch %	Lines
crates/polars-io/src/path_utils/hugging_face.rs	46.66%	136 Missing ⚠️
crates/polars-io/src/path_utils/mod.rs	83.33%	6 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #17665      +/-   ##
==========================================
- Coverage   80.72%   80.71%   -0.02%     
==========================================
  Files        1485     1486       +1     
  Lines      195527   195798     +271     
  Branches     2782     2782              
==========================================
+ Hits       157839   158029     +190     
- Misses      37176    37257      +81     
  Partials      512      512              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[c-peters]

Nice!

[ritchie46]

Ran a benchmark:

# Download full parquet:
- Polars: 12,1s
- DuckDB: 2min 14s
- pandas: 13.6s
- dask: 13.6s

# Download with a limit (so download must stop early)
- Polars <1s
- DuckDB: 1,5s
- pandas: 12,4s
- dask: 13.9s