Audit all data flows: eliminate default/empty values that cause downstream failures

[STRRL]

Description

Multiple dashboard bugs (#68, #69) stem from the same root pattern: data fields are left empty/default at write time, then downstream code assumes they will be populated, leading to incorrect display or crashes.

Known instances

1. InsertCompletedJob writes empty runner_name and runner_set_name — frontend crashes on undefined.split()
2. RecordJobCompleted(jobID, result) doesn't pass through runnerName/runnerSetName even though the caller has them
3. Job result stored in DB but toProtoJobResult() returns UNSPECIFIED (enum 0) for unrecognized values — protobuf omits the field — frontend defaults completed jobs to failure
4. Frontend fetchRecentJobs fallback: j.completedAt ? 'failure' : 'running' — any unknown result on a completed job becomes failure

What needs to happen

• Audit every write path (job start, job complete, runner state transitions) — ensure all required fields are populated at the point of write, not left for downstream to guess
• Audit every read/display path — ensure no code assumes a field is always present without a guard
• Specifically trace the full lifecycle: Scale Set API event → controller → JobStore → API server (protobuf) → frontend fetcher → UI component
• For each field (runnerName, runnerSetName, result, startedAt, completedAt), verify it is correctly set at every stage
• Eliminate patterns where a field is "expected but empty" — either make it required and always populated, or make it explicitly optional and handle the missing case everywhere

Related issues

• Bug: job result tracking incorrect — completed jobs shown as FAIL, COMPLETED counter stays at 0 #68 — COMPLETED counter always 0, all jobs show FAIL
• Bug: dashboard crashes with black screen (React runtime error) #69 — Dashboard black screen crash (undefined.split)