fix(api): add structured logs for sandbox delete 500/404 paths

[AdaAibaby]

Fixes #3529

Problem

DeleteSandboxesSandboxID has three logging gaps:

1. ErrSandboxOperationFailed → 500 with no log or telemetry — the most common 500 cause (node unreachable, gRPC error) emits nothing to Loki and records no span error.
2. deleteSnapshot failure message is ambiguous — "error deleting sandbox" does not distinguish kill failure from DB failure.
3. 404 logged at Debug level, no teamID — suppressed in production; 404 frequency per team is invisible.

Changes

packages/api/internal/handlers/sandbox_kill.go:

Branch	Before	After
ErrSandboxOperationFailed	no log	Error + sandbox_id + team_id + zap.Error + telemetry.ReportError
deleteSnapshot != nil	telemetry.ReportError only	Error log with distinct message "Failed to delete sandbox snapshot from DB"
default unexpected error	telemetry.ReportError only	Error log added
ErrSandboxNotFound	Debug	Info + team_id
final 404	Debug, no team_id	Info + team_id, message "not running and no snapshot"

Why this is not a duplicate

Searched open PRs for sandbox delete log, sandbox kill log, sandbox 500 log — no existing PR covers this.

Test plan

• go build ./packages/api/internal/handlers/ passes
• go vet ./packages/api/... passes
• Manually trigger a 500 (sandbox on unreachable node) → confirm "Failed to kill sandbox on orchestrator node" in Loki with sandbox_id and team_id
• Manually trigger a 404 (non-existent sandbox) → confirm "Sandbox not found: not running and no snapshot" at Info level

/cc @jakubno @dobrac @ValentaTomas @arkamar @tvi @tomassrnka Looking forward to your code review.