CBG-4746: Deduplicate channel set outside processEntry

[gregns1]

CBG-4746

• Changed delay slightly in tool to allow toll to run faster
• Switch to return list of channel ID not maps from processEntry then deduplicate the channel IDs outside processEntry mutex

Before:

After:

Pre-review checklist

• Removed debug logging (fmt.Print, log.Print, ...)
• Logging sensitive data? Make sure it's tagged (e.g. base.UD(docID), base.MD(dbName))
• Updated relevant information in the API specifications (such as endpoint descriptions, schemas, ...) in docs/api

Dependencies (if applicable)

• Link upstream PRs
• Update Go module dependencies when merged

Integration Tests

• GSI=true,xattrs=true https://jenkins.sgwdev.com/job/SyncGatewayIntegration/153/

[Copilot]

Pull Request Overview

This PR optimizes channel cache performance by changing the return type of processEntry from a channel set to a slice of channel IDs, and moving channel deduplication outside the mutex-protected section. This reduces contention and improves performance for the cache performance tool.

• Changed AddToCache and processEntry methods to return []channels.ID instead of channels.Set
• Updated all callers to convert the slice to a set using channels.SetFromArrayNoValidate for deduplication
• Reduced delay in cache performance tool from 500ms to 100ms for faster testing

Reviewed Changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.

File	Description
tools/cache_perf_tool/dcpDataGeneration.go	Reduced vBucket creation delay from 500ms to 100ms
db/channel_cache.go	Changed AddToCache return type and implementation to use slice instead of set
db/change_cache.go	Updated processEntry and related methods to return slices, with deduplication moved outside mutex

[torcolvin]

Concept seems fine, I think a quick look to see if we can hide the duplication in our our functions is reasonable, and we should definitely document which functions which could return duplicate channels in output.