Avoid full materialization of levels in failing MultiIndex validations

[amerberg]

This is a followup to #2118, which allowed us to avoid materializing full level values of MultiIndex levels for successful validation. That PR, however, fell back to full materialization upon hitting a schema error, to ensure that failure cases would be reported correctly. This PR extends the benefit of the optimization to failing validations by joining the failure_cases from the unique values to a lookup table to build the failure_cases for the full values.

This should be particularly beneficial for running validations that are expected to fail, such as validating against a union DataFrame[Schema1] | DataFrame[Schema2] | DataFrame[Schema3] where any given dataframe is only expected to conform to one of the schemas.

[codecov]

Codecov Report

❌ Patch coverage is 97.05882% with 1 line in your changes missing coverage. Please review.
✅ Project coverage is 83.90%. Comparing base (f5c55cb) to head (f97808d).
⚠️ Report is 9 commits behind head on main.

Files with missing lines	Patch %	Lines
pandera/backends/pandas/components.py	97.05%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2187      +/-   ##
==========================================
+ Coverage   81.51%   83.90%   +2.38%     
==========================================
  Files         137      137              
  Lines       10898    10631     -267     
==========================================
+ Hits         8884     8920      +36     
+ Misses       2014     1711     -303     

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

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[cosmicBboy]

thanks @amerberg !