swss-common-bridge: replay unchanged producer updates

[BYGX-wcr]

Description of PR

Summary:
The producer bridge cached the last update and skipped identical requests. That cache can outlive a DPU control-plane reset, so an HA_SET entry removed from the destination DB was not restored when hamgrd replayed unchanged data. Always apply valid producer requests so replay can recover externally lost DB state. Consumer-side caching remains unchanged.

Fixes: sonic-net/sonic-buildimage#28073

Type of change

• Bug fix
• New feature
• Refactor / cleanup
• Documentation update
• Test improvement

Approach

What is the motivation for this PR?

A DPU control-plane reset can clear DASH_HA_SET_TABLE while swss-common-bridge remains running. Replaying the same HA_SET data previously returned "No change in data" without writing it, leaving the destination DB entry missing.

How did you do it?

• Removed producer-side TableCache deduplication.
• Apply every valid Set or Del request to the ProducerTable.
• Kept TableCache for consumer-side incremental merge and duplicate-notification filtering.
• Updated duplicate and ZMQ consumer-recreation tests to require identical updates to be replayed.

How did you verify/test it?

• cargo fmt --check --all
• cargo build --workspace --all-features
• cargo test -p swss-common-bridge (9 passed)
• Built the Debian runtime and debug-symbol packages.
• Installed and verified the packages in dash-hadpu0 through dash-hadpu3 on two SmartSwitches.
• Completed config reload on both switches and verified all eight hamgrd processes were running.

Any platform specific information?

Validated on SONiC SmartSwitch DPU containers using both Redis ProducerStateTable and ZmqProducerStateTable paths.

Documentation

No documentation change is required; this restores producer replay correctness without changing the public interface.

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines:
There may be pipelines that require an authorized user to comment /azp run to run.

[azure-pipelines]

Azure Pipelines:
Successfully started running 1 pipeline(s).

[Copilot]

Pull request overview

This PR fixes a producer-side replay correctness issue in swss-common-bridge where identical producer updates were previously deduplicated and skipped. Because that dedup cache could outlive an external DB state loss (e.g., DPU control-plane reset clearing DASH_HA_SET_TABLE), hamgrd replays of unchanged data would not restore missing entries; the bridge now always applies valid producer requests so external state loss can be recovered.

Changes:

• Removed producer-side TableCache deduplication so identical SET/DEL requests are always applied to the ProducerTable.
• Updated tests to require identical updates to be replayed (including after ZMQ consumer recreation).
• Narrowed TableCache’s documented purpose to consumer-side merge + duplicate-notification filtering and removed the producer-oriented replace_kfv API.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
crates/swss-common-bridge/src/producer.rs	Always applies producer requests (no dedup) and updates tests to assert identical updates are replayed.
crates/swss-common-bridge/src/lib.rs	Updates TableCache docs to consumer-only semantics and removes producer-side replace_kfv dedup method.