Preserve duplicate Cassandra list values

[dipeshbabu]

What

Include the monotonic timestamp in the Cassandra clustering key, keep set inserts idempotent with a fixed timestamp, validate existing table schemas, and add duplicate/removal integration regressions.

Why

The previous primary key used only key and value, so repeated ordered values overwrote each other. An empty MinHash repeats the same band hash, which meant its key metadata lost bands and removal left stale bucket entries.

Impact

Cassandra ordered storage now preserves every inserted value while unordered bucket storage remains deduplicated. Tables created with the old primary key fail with a clear rebuild instruction instead of silently retaining unsafe semantics.

Root cause

The ordering timestamp was a regular column rather than part of row identity.

Checks

• uv run pytest -q test/test_lsh.py (24 passed)
• uv run pytest -q test/test_integration.py (36 skipped because backend services are unavailable locally; Cassandra CI runs them)
• uv run ruff check datasketch/storage.py test/test_integration.py
• uvx pyright (0 errors)
• git diff --check

[dipeshbabu]

Superseded by upstream pull request ekzhu#327.