fix: route workflow IDs and retry jitter through platform seams

[DABH]

Describe the bug

google.adk.platform provides injectable time and uuid providers so that embedders can customize how generated values are produced - for example to make agent runs deterministic and reproducible for testing, simulation, or auditing. Event IDs, invocation IDs, timestamps, and client function-call IDs already route through them, but three call sites on the workflow HITL/retry path still call the stdlib directly, so they bypass any installed provider:

1. events/request_input.py - RequestInput.interrupt_id defaults to uuid.uuid4(). This is the ID that recorded user FunctionResponses reference on resume; an ID minted outside the installed provider can't be reproduced, which breaks interrupt matching.
2. workflow/_tool_node.py - ToolContext.function_call_id is minted with uuid.uuid4() instead of platform_uuid.new_uuid() (which flows/llm_flows/functions.py already uses for the same purpose).
3. workflow/utils/_retry_utils.py - retry backoff jitter draws from the global random module. Jitter defaults to 1.0 for any RetryConfig, so every retried node computes an unreproducible delay.

Describe the fix

• Add a platform random seam - google.adk.platform._random with get_random() / set_random_provider() / reset_random_provider(), exported from google.adk.platform per the visibility convention - mirroring the existing time/uuid providers (ContextVar-based, default = stdlib random.Random()).
• Route the three call sites through the platform seams. Default behavior is unchanged (same uuid4/uniform semantics when no provider is installed).
• Migrate the two existing jitter tests from mock.patch('random.uniform', ...) to injecting a mock via set_random_provider, exercising the new seam with the exact same assertions.

This follows the injectable-provider approach also adopted in google/adk-java#1255.

Testing plan

New unit tests:

• tests/unittests/platform/test_random.py - default provider, custom provider, reset (mirrors test_uuid.py).
• tests/unittests/events/test_request_input.py - default interrupt_id is a uuid; minted via the platform id provider; explicit id preserved.
• tests/unittests/workflow/test_tool_node.py::test_tool_node_function_call_id_uses_platform_id_provider - the tool receives a provider-minted function_call_id.
• tests/unittests/workflow/utils/test_retry_utils.py::test_jitter_uses_platform_random_provider - a seeded provider makes the computed delay reproducible.

$ pytest tests/unittests/workflow/ tests/unittests/platform/ tests/unittests/events/
791 passed, 11 skipped, 10 xfailed in 6.95s

All pre-commit hooks (isort, pyink, addlicense, new-file-prefix, ADK compliance, codespell) pass.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[DABH]

@DeanChensj The prior CI failure appears to just be because my PR was out of date with the main branch. I've updated and we should be good to go. Would you mind re-approving the CI run so things can retry? Thanks again!!

[DABH]

Ah, looks like main is broken on CI now - once main is green, I'll update here one more time and hopefully all will go smoothly (I've verified locally and it looks green basing this off an older commit from main.). Keeping an eye out... thanks again!