Use xxHash64 for event_instance_id to prevent hash collisions

[tombonfert]

Problem

event_instance_id was generated with Spark's crc32, whose 32-bit output space produces collisions in practice. A user hit a real case where two distinct instances of the same event on one container (same event_name, different time intervals) hashed to the same id.

Because event_instance_id is a MERGE key for event_instance_fact and the join key from stats_aggregator_fact and point_value_fact, a collision silently drops or mis-associates event instances in the gold layer.

Fix

Switch event_instance_id generation from crc32 to xxhash64 (64-bit), widening the key space enough to keep it collision-free at scale. The change is made in the single shared helper (generate_event_instance_id_column), so event_instance_fact and the aggregation facts stay in lockstep automatically.

Changes

• event_instance_util.py: crc32 to xxhash64 for both the container-event and timestamp-based branches; docstring updated.
• fact_schema.py: event_instance_id typed as LongType to match the 64-bit hash and STATS_AGGREGATOR_FACT_SCHEMA (cosmetic; persistence projects by name).
• Tests and a docstring in stats_aggregator.py updated to mirror the new hash.

Notes

• event_instance_id values are now 64-bit and can be negative. Equality still holds for MERGE and joins, so correctness is unaffected, but any downstream consumer assuming a positive or 32-bit id should be aware.
• No gold-table migration. Existing rows keep their old ids; affected instances are re-inserted with new ids on the next run.

Verification

Targeted unit tests, the events/aggregations/incremental sweep, and the statistics integration tests pass; make lint clean.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 89.01%. Comparing base (23541ff) to head (425fa9e).

Additional details and impacted files

@@           Coverage Diff           @@
##             main      #85   +/-   ##
=======================================
  Coverage   89.01%   89.01%           
=======================================
  Files          61       61           
  Lines        5207     5207           
  Branches      628      628           
=======================================
  Hits         4635     4635           
  Misses        461      461           
  Partials      111      111           

Flag	Coverage Δ
query_engine	84.77% <ø> (ø)
reporting	94.24% <100.00%> (ø)

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

Files with missing lines	Coverage Δ
...impulse_reporting/aggregations/stats_aggregator.py	94.25% <ø> (ø)
src/impulse_reporting/persist/fact_schema.py	100.00% <ø> (ø)
src/impulse_reporting/util/event_instance_util.py	100.00% <100.00%> (ø)

🚀 New features to boost your workflow:

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