chore(store): remove dead clock param + fix broken main (transactionRunner in two purger tests)

[endrju19]

Two related changes in one PR. The first is an urgent fix — main currently does not compile its tests.

1. fix: repair broken main (commit 1)

./gradlew :okapi-core:compileTestKotlin fails on main right now:

OutboxPurgerTest.kt:157  No value passed for parameter 'transactionRunner'.
OutboxPurgerTest.kt:196  No value passed for parameter 'transactionRunner'.

Cause — a semantic merge collision between #55 and #54:

• fix(purger): preserve partial-batch progress in error log on mid-loop failure #55 added two OutboxPurger tests while transactionRunner was still nullable, so they didn't pass it.
• feat(core): require TransactionRunner in OutboxScheduler/OutboxPurger (breaking) #54 (merged separately) made transactionRunner required and updated every call site it could see — but it never saw fix(purger): preserve partial-batch progress in error log on mid-loop failure #55's two new tests (different unmerged branch).
• Each PR was green against its own base. Git merged both cleanly (different lines → no textual conflict), but the combination violates feat(core): require TransactionRunner in OutboxScheduler/OutboxPurger (breaking) #54's new required-parameter contract. No CI run ever compiled fix(purger): preserve partial-batch progress in error log on mid-loop failure #55's tests against feat(core): require TransactionRunner in OutboxScheduler/OutboxPurger (breaking) #54's signature.

This would have been caught by GitHub's "Require branches to be up to date before merging" (or a merge queue) — worth enabling.

Fix: add the missing transactionRunner = noOpTransactionRunner() to both call sites (the helper already exists in the file and is used by the other 7 purger tests).

2. chore: remove dead clock from the stores (commit 2)

clock became unused in PostgresOutboxStore / MysqlOutboxStore after #58 — that PR replaced the findOldestCreatedAt per-status clock.instant() pre-seed with an empty map, so the stores no longer read a clock at all. (clock was in the stores only to feed that pre-seed, since the very first commit.)

• Removed the clock constructor parameter + java.time.Clock import from both adapters.
• Dropped the now-pointless clock pass-through from the two okapi-spring-boot store bean factories (outboxStore()).
• Updated all store construction sites (benchmarks + integration tests) to the single-arg form PostgresOutboxStore(connectionProvider).
• Left untouched every clock that still feeds OutboxPublisher / OutboxEntryProcessor / OutboxProcessor / MicrometerOutboxMetrics — those components legitimately take a clock. (Verified no stale clock reference, import, comment, or name remains.)

Breaking change (documented in CHANGELOG): only affects direct constructor users who passed an explicit clock — PostgresOutboxStore(connectionProvider, clock) must drop the second arg. The usual PostgresOutboxStore(connectionProvider) form is unchanged; Spring Boot users are unaffected.

Test plan

• ./gradlew testClasses — all main + test + jmh sources compile (was failing on main)
• ./gradlew ktlintCheck — green (no orphaned imports)
• ./gradlew :okapi-core:test :okapi-spring-boot:test — green (purger tests + autoconfig wiring)
• ./gradlew :okapi-integration-tests:test --tests "*PostgresOutboxStoreTest" --tests "*MysqlOutboxStoreTest" — green (both adapters against real Postgres + MySQL via Testcontainers)