Skip queued callbacks after a group error

[bensynapse]

I run Live Tennis API.

With NewErrSizedGroup(1, TermOnErr), a callback waiting for capacity currently runs even after the active callback returns an error.
The termination check now runs after semaphore acquisition, so queued callbacks see errors recorded while they were waiting.
Already running callbacks can still finish.

The regression reproduces the bug before the fix.
It covers both locking modes and execution without TermOnErr, and checks that skipped callbacks release their permits.

Verified on Go 1.23.4 with TZ=America/Chicago go test -timeout=60s -race -covermode=atomic -coverprofile=profile.cov ./....
The suite passes with 99.1% statement coverage. The new test passes 50 race runs.
go build -race ./... and go vet ./... pass. golangci-lint run --timeout=3m reports zero issues.

[umputun]

lgtm

[coveralls]

Coverage Report for CI Build 37111638545

Coverage remained the same at 98.87%

Details

• Coverage remained the same as the base build.
• Patch coverage: 3 of 3 lines across 1 file are fully covered (100%).
• No coverage regressions found.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

No coverage regressions found.

---

Coverage Stats

Relevant Lines:	177
Covered Lines:	175
Line Coverage:	98.87%
Coverage Strength:	1955.24 hits per line

---

💛 - Coveralls