fix: link_parent more realible and delete topic partitions

[fraidev]

The fix of #4094 was a wrong fix to have sure that all partitions are childrens of its Topic.

I reverted this PR, and I made the link_parent and unlink_parent more reliable.
Because they were being called correct, but link_parent it was using stale objects.

Now they are getting the parent from the store to have sure to not lose data.

How to test:

fluvio topic create data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic delete data
fluvio partition list

Must output no results.

This should also fix logs like these after each 5min (SC_RECONCILIATION_INTERVAL_SEC):

 sc        | 2024-10-25T21:16:09.364724Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 11, proposed: 10
sc        | 2024-10-25T21:16:09.366955Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 5, proposed: 4
sc        | 2024-10-25T21:16:09.370491Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 14, proposed: 13
sc        | 2024-10-25T21:16:09.378256Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 8, proposed: 7
sc        | 2024-10-25T21:16:09.379879Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 8, proposed: 7
sc        | 2024-10-25T21:16:09.381140Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 6, proposed: 5
sc        | 2024-10-25T21:16:09.382941Z ERROR MetadataDispatcher{spec="Partition" namespace="default"}:process_ws_action: fluvio_stream_dispatcher::dispatcher::metadata: error: Partition, applying attempt to update by stale value: current version: 12, proposed: 11

[sehz]

Fix seems to look good. Can you add CLI Test as described in the issue?

fluvio topic create data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic delete data
fluvio partition list

[fraidev]

Fix seems to look good. Can you add CLI Test as described in the issue?

fluvio topic create data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic add-partition data
fluvio topic delete data
fluvio partition list

I created one at #4094 at tests/cli/fluvio_smoke_tests/add-partitions.bats.

[sehz]

But then that didn't catch error. Can you create PR to reproduce bug?

[fraidev]

But then that didn't catch error. Can you create PR to reproduce bug?

The #4094 fixed this bug with changes at crates/fluvio-sc/src/stores/topic/store.rs. (In a wrong way)
I reverted these changes there, and fixed it with changes in crates/fluvio-stream-dispatcher/src/metadata/local.rs

I can create a PR with only crates/fluvio-sc/src/stores/topic/store.rs reverted, this test should catch it than. What do you think?

[sehz]

Let's not revert anything. This PR should be suffice for fix but we should make sure we can reproduce and confirm fix is applied. Since this bug can be manually reproduced, let's create PR to reproduce the bug (test should confirm bug). Once that is in, this PR should update test to assert bug fix is working

[sehz]

LGTM