Implement support for retaining Pandas index

[philippjfr]

The Pandas support in HoloViews has always been severely hampered by the fact that we do not properly support indexes, i.e. if a user references an index column we are forced to call .reset_index() which is inefficient and breaks one core principle of HoloViews, which is that we simply provide thin wrappers around data. It also has significant performance implications both from a memory perspective and from a speed perspective since indexes can provide significant speedups when indexing or performing aggregations. One other major issue solved by supporting indexes is the problem of having multiple elements providing a view onto the same DataFrame, e.g. when you have an NdOverlay of curves each visualizing a different column.

Implements #2537

• Implement all operations correctly
  • __init__
  • validate
  • dtype
  • dimension_type
  • range
  • iloc (check multi-index)
  • values (check multi-index)
  • sort
  • reindex
  • sample
• Optimized operations
  • select
  • groupby
• Check operations that use dataset.data directly do not make (now invalid) assumptions about indexes
• Add tests

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 20.06579% with 243 lines in your changes are missing coverage. Please review.

Project coverage is 26.96%. Comparing base (342d81c) to head (4546212).
Report is 15 commits behind head on main.

❗ Current head 4546212 differs from pull request most recent head 1476fac. Consider uploading reports for the commit 1476fac to get more accurate results

Files	Patch %	Lines
holoviews/tests/core/data/test_pandasinterface.py	21.17%	134 Missing ⚠️
holoviews/core/data/pandas.py	16.39%	102 Missing ⚠️
holoviews/core/data/ibis.py	41.66%	7 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main    #6061       +/-   ##
===========================================
- Coverage   88.68%   26.96%   -61.73%     
===========================================
  Files         316      318        +2     
  Lines       66072    67132     +1060     
===========================================
- Hits        58598    18104    -40494     
- Misses       7474    49028    +41554     

Flag	Coverage Δ
ui-tests	26.96% <20.06%> (+3.17%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

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

[droumis]

may partially resolve #6058

[philippjfr]

Looks good, left a few suggestions and questions for you.

[philippjfr]

Okay, I'm happy with this PR personally. We really need to merge this and test it extensively in all kinds of scenarios over the next few weeks.