Add grouped validation summary to stderr output

[DanielSebagala]

Adds a ValidationSummaryHandler to validateData.py that prints a categorized summary of all errors and warnings at the end of validation.

Problem

The current validator streams messages chronologically as they're found. For large studies with many issues, curators have to scroll through the full log to understand what went wrong and how often. The per-file summaries (line count with warnings/errors) help, but don't aggregate across the full study or group by message type.

Solution

A new ValidationSummaryHandler (a standard logging.Handler subclass) collects all ERROR and WARNING messages during validation and prints a grouped, counted summary to stderr after validation completes:

------------------------------------------------------------
VALIDATION SUMMARY
------------------------------------------------------------
ERRORS (8):
  [4] Normal sample id not in list of sample ids...
  [2] Value of numeric attribute is not a real number
  [1] No case list found with stable_id 'brca_tcga_pub_all'
  [1] datatype definition for attribute 'DFS_MONTHS' must be NUMBER

WARNINGS (10):
  [4] Unrecognized field in meta file
  [1] Missing clinical data for a patient associated with samples
  [1] Given value for Variant_Classification column is not one of the expected values
------------------------------------------------------------

Changes

• New ValidationSummaryHandler class (~50 lines) after MaxLevelTrackingHandler
• Two lines in main_validate() to register the handler
• Summary flush and print before returning exit status

Testing

• Tested against study_various_issues, study_es_3, study_wr_clin — all produce correct grouped summaries
• 156 existing unit tests pass (2 pre-existing failures unrelated to this change)
• No changes to existing logging behavior — the summary is additive, printed to stderr alongside the existing Validation of data {status} line