feat(filegen): add %EPOCH_S/MS/US/NS Unix epoch time directives (PIPE-1408)

[Dylan-M]

Proposed Change

The data library's time-format directives had no Unix epoch token. %s is already the sub-second token, so it can't carry strftime's epoch-seconds meaning. This adds four unit-encoded directives: %EPOCH_S, %EPOCH_MS, %EPOCH_US, %EPOCH_NS.

Epoch values are integers the ctime layout formatter can't express, and ctime rejects any format containing digits. So they're substituted in processTimestamps before the line reaches ctime, from the same now the other directives use. %s and every other directive are left untouched.

How the reviewer can validate

Note: the manual validation procedure below is an AI-generated script. Review it before running.

Manual:

1. make build
2. printf 'ts_s=%%EPOCH_S ts_ms=%%EPOCH_MS ts_us=%%EPOCH_US ts_ns=%%EPOCH_NS iso=%%Y-%%m-%%dT%%H:%%M:%%SZ\n' > /tmp/epoch.log
3. ./blitz --generator-type filegen --generator-filegen-source /tmp/epoch.log --generator-filegen-rate 300ms --output-type stdout
4. Confirm each token becomes an integer of the right width (10/13/16/19 digits), all from one instant. %Y-… should still format normally. Example: ts_s=1786630895 ts_ms=1786630895215 ts_us=1786630895215112 ts_ns=1786630895215112371.

Automated: go test ./generator/filegen/. TestReplaceEpochDirectives asserts the exact integer per unit and integer form; TestReplaceEpochDirectives_LeavesSubSecondUntouched guards %s; the TestTimestampProcessing epoch case covers end-to-end substitution.

Checklist

• Changes are tested
• CI has passed

[Dylan-M]

Merge activity

• Aug 13, 8:52 PM UTC: A user started a stack merge that includes this pull request via Graphite.
• Aug 13, 8:53 PM UTC: @Dylan-M added this pull request to the GitHub merge queue with Graphite.