chore(tests): normalize recording IDs and timestamps to reduce git diff noise (#3676)

IDs are now deterministic hashes based on request content, and
timestamps are normalized to constants, eliminating spurious changes
when re-recording tests.

## Changes
- Updated `inference_recorder.py` to normalize IDs and timestamps during
recording
- Added `scripts/normalize_recordings.py` utility to re-normalize
existing recordings
- Created documentation in `tests/integration/recordings/README.md`
- Normalized 350 existing recording files

tests/integration/recordings/responses/f1592dee71e5.json

@@ -30,15 +30,15 @@
       "__type__": "ollama._types.ChatResponse",
       "__data__": {
         "model": "llama3.2-vision:11b",
-        "created_at": "2025-09-03T17:54:32.086616Z",
+        "created_at": "1970-01-01T00:00:00.000000Z",
         "done": true,
         "done_reason": "stop",
-        "total_duration": 3537246333,
-        "load_duration": 130547125,
+        "total_duration": 0,
+        "load_duration": 0,
         "prompt_eval_count": 18,
-        "prompt_eval_duration": 140216250,
+        "prompt_eval_duration": 0,
         "eval_count": 56,
-        "eval_duration": 3262609875,
+        "eval_duration": 0,
         "message": {
           "role": "assistant",
           "content": "The image is of a golden retriever puppy. The puppy is looking directly at the camera with its mouth open and tongue out. The puppy is white with golden ears and a black nose. The background is out of focus, but it appears to be a grassy field.",