feat: issue #5674- improve wf parsing

[srivatsa-av28]

Closes #5674

📑 Description

Added an in-process parse cache to WorkflowStore keyed on
(tenant_id, workflow_id), using the workflow's existing revision
field as the cache validity key.

• Cache hit: returns the previously parsed Workflow object directly,
  skipping cyaml.safe_load() and Parser.parse() entirely
• Cache miss: parses as normal and stores the result
• Cache invalidation: invalidate_workflow_cache() is called from
  delete_workflow() to evict stale entries on deletion
• Thread safety: a threading.Lock protects the cache dict
  (safe for the ThreadPoolExecutor used in WorkflowScheduler)

Changes

• keep/workflowmanager/workflowstore.py
  • Added _CachedWorkflow dataclass to hold parsed workflow + revision
  • Added _parse_cache dict and _cache_lock to __init__
  • Updated get_workflow() to check cache before parsing
  • Added invalidate_workflow_cache() method
  • Called invalidate_workflow_cache() from delete_workflow()

Notes

• No schema migration required
• No behaviour change on the hot path — cache is purely additive
• Fully backwards compatible
• revision is already incremented by add_or_update_workflow() on
  every update, making it a reliable and zero-cost cache key

✅ Checks

• My pull request adheres to the code style of this project
• My code requires changes to the documentation
• I have updated the documentation as required
• All the tests have passed

ℹ Additional Information

[CLAassistant]

All committers have signed the CLA.