fix: jobs report success during dependency outages

[silent-cipher]

Problem

data_retention_poll recorded jobs_completed_total{handler_result="success"} on every run, even when the check couldn't execute because the PDP subgraph was down. A subgraph outage silently took down mainnet + calibration data-retention checks for ~6 days with no alarm, stalling SP approval on insufficient dataSetChallengeStatus samples. providers_refresh had the same latent bug: it discarded the boolean from loadProviders() and hardcoded success.

A job's success should mean the check was able to run, so a dependency outage must be recorded as a job failure.

Changes

• data-retention.service.ts: pollDataRetention() now fails the job (throws DataRetentionDependencyError) when a check dependency is unavailable - missing subgraph endpoint, baseline-load failure, subgraph meta fetch failure, or a subgraph query failure. Transient per-provider failures and an empty-but-healthy provider set stay success. On a batch query failure the poll still records healthy batches before failing, and a failure is logged once.
• wallet-sdk.service.ts: loadProviders() returns its success boolean instead of discarding it.
• jobs.service.ts: handleProvidersRefreshJob fails the job when loadProviders() reports failure.

These propagate through recordJobExecution, which records handler_result="error".

For #591

This PR doesn't closes it because of the below remaining task:

• The job-completion alarm fires on a sustained dependency outage, including when only error results exist.

[Copilot]

Pull request overview

This PR fixes job result reporting so dependency outages (e.g., PDP subgraph down, baseline DB load failure, on-chain provider registry load failure) are surfaced as job failures instead of being recorded as successful completions, ensuring jobs_completed_total{handler_result="error"} reflects real execution failures.

Changes:

• Make WalletSdkService.loadProviders() return a success boolean and use it in the providers_refresh job to fail the job when on-chain provider loading fails.
• Update DataRetentionService.pollDataRetention() to throw a dependency-specific error when critical dependencies are unavailable, including on batch subgraph query failures.
• Expand data-retention.service.spec.ts coverage to assert job-failure behavior for dependency outages while preserving “success” for empty-but-healthy provider sets and transient per-provider processing failures.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
apps/backend/src/wallet-sdk/wallet-sdk.service.ts	Return a boolean from loadProviders() so callers can treat dependency failures as job failures.
apps/backend/src/jobs/jobs.service.ts	Fail providers_refresh job when loadProviders() reports failure (so handler_result becomes error).
apps/backend/src/data-retention/data-retention.service.ts	Throw DataRetentionDependencyError for missing endpoint / baseline load failure / subgraph meta & query failures; fail after recording partial batch results.
apps/backend/src/data-retention/data-retention.service.spec.ts	Add/update tests to verify dependency outages now fail the job while partial/transient failures remain success.

[Copilot]

Pull request overview

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.