[Stability] Outbound Easebuzz HTTP calls have no timeout/retry controls

[chromonav]

Branch

• dev

Evidence

• easebuzz/easebuzz/utils/payment.py#L541
• easebuzz/easebuzz/utils/transaction.py#L379
• easebuzz/easebuzz/utils/transaction_date.py#L299
• easebuzz/easebuzz/utils/refund.py#L390
• easebuzz/easebuzz/utils/payout.py#L345

All these paths call requests.post(...) without timeout and without retry/backoff strategy.

Problem

External gateway calls run without network time bounds. A slow/downstream gateway can block worker threads and degrade request throughput.

Risk/Impact

• Stuck web workers/background jobs under network stalls.
• Cascading latency spikes during gateway degradation.
• Poor operational recoverability (no retry/backoff policy).

Replication suggestion

1. Simulate gateway endpoint slowness (iptables/tc proxy delay or blackhole).
2. Trigger payment/transaction/refund/payout APIs.
3. Observe requests hanging until OS/network defaults, with no controlled timeout.

Expected: bounded timeout and controlled retry policy.
Actual: unbounded waits.

Resolution suggestion

• Add explicit timeout tuple: timeout=(connect_timeout, read_timeout) for every requests.post.
• Implement retry with capped exponential backoff for transient network failures (idempotent operations only).
• Normalize error handling to return typed failure payloads and structured logs.

Acceptance criteria

• Every outbound HTTP call to Easebuzz defines explicit timeout.
• Transient failures are retried per policy; permanent failures fail fast.
• Observability includes failure reason and latency metrics.
• Integration tests cover timeout behavior.