Fix: route the last two core/src console.error calls through the shared logger

[AmaadMartin]

Please ensure you have read the contribution guide before creating a pull request.

Link to Issue or Description of Change

1. Link to an existing issue (if applicable):
   N/A
2. Or, if no issue exists, describe the change:
   Problem: Two sites in core/src write their failure diagnostics with console.error: the summarizer catch in agent_controlled_context_compactor.ts and the runAsync catch in load_memory_tool.ts. A consumer who calls setLogger(), setLogger(null) or setLogLevel() cannot silence or redirect either line. The records also miss the ADK label, level and timestamp that every other ADK diagnostic carries.

Solution: Both sites now call logger.error from core/src/utils/logger.ts, so only the identifier changes. I used the logger facade rather than a module-level const logger = getLogger(), because the facade reads currentLogger at call time and so still honours a later setLogger(). The caught error stays a second argument, which keeps the Error object intact for a custom logger. Behaviour is otherwise unchanged: the compactor still swallows the error and clears its flags, and LoadMemoryTool still rethrows.

The ERROR in prefix in the load-memory message is dropped because the logger already renders the level, so keeping it produced ERROR: [ADK] <ts> ERROR in LoadMemoryTool runAsync: ....

Collision check: I listed the 400 open PRs on this fork and searched every non-main commit for edits to these two files and to the two message strings. Four PRs change core/src/utils/logger.ts (#807, #795, #697, #683) and two touch other parts of the compactor (refactor/compaction-utils, feat/apply-rewinds-live-event-filter). None converts these two call sites, so this branches from main.

Testing Plan

Please describe the tests that you ran to verify your changes. This is required for all PRs that are not small documentation or typo fixes.
Unit Tests:
[x] I have added or updated unit tests for my change.
[x] All unit tests pass locally.

One new test per changed statement. Existing tests are untouched.

• core/test/context/agent_controlled_context_compactor_test.ts — "logs the summarizer failure"
• core/test/tools/load_memory_tool_test.ts — "logs the failure before rethrowing"

Both spy on getLogger().error, which the facade forwards to, and both assert on the same Error instance the code under test threw. The load-memory test builds a real Context over an InvocationContext whose memoryService rejects, so it drives the real Context.searchMemory path.

Proof each test can fail. I restored the original console.error line and re-ran the matching test:

• compactor: AssertionError: expected "error" to be called with arguments: [ 'Compaction failed:', …(1) ] / Number of calls: 0
• load memory: AssertionError: expected "error" to be called with arguments: [ …(2) ] / Number of calls: 0

Commands run on the pushed commit:

npx vitest run --project unit:core core/test/context/agent_controlled_context_compactor_test.ts core/test/tools/load_memory_tool_test.ts   # 13 passed
npx vitest run --project unit:core                                                                                                        # 201 files, 2733 passed
npm run build --workspace core                                                                                                            # ok
npm run lint                                                                                                                              # clean
npm run format:check                                                                                                                      # clean
npm run ts:check                                                                                                                          # pre-existing errors only; none in the four files I touched

Manual End-to-End (E2E) Tests:
Call setLogger(null) from @google/adk, then drive LoadMemoryTool.runAsync into its catch with a memory service that rejects. Before this change a bare line prints on the console. After it, nothing prints.

Checklist

[x] I have read the CONTRIBUTING.md document.
[x] I have performed a self-review of my own code.
[x] I have commented my code, particularly in hard-to-understand areas.
[x] I have added tests that prove my fix is effective or that my feature works.
[x] New and existing unit tests pass locally with my changes.