[CMGR-76557] fix: reconnect tail-log stream after EOF or transient error

[GeorgeAlexandru]

Summary

• Adds a resilient while(true) reconnect loop in TailLog.tailLog() so the CLI never exits when the log stream ends
• Swallows transient errors (no error.code: 404, network drops, Azure SAS URL expiry) and retries after a 2-second delay
• Re-throws fatal CLI-coded errors (error.code set: MAX_RETRY_REACHED, NO_IMS_CONTEXT) so auth and config failures still surface correctly
• Ctrl+C still exits immediately (SIGINT kills the process regardless of the loop)

Jira: https://jira.corp.adobe.com/browse/CMGR-76557
Root cause: executeWithRetries only retried on HTTP 401/403. EOF at midnight log rotation resolved normally — no error, no retry, silent CLI exit.

Test plan

• New test: tail-log - reconnects after normal stream end — verifies sdk.tailLog called ≥2 times after EOF
• New test: tail-log - retries silently on transient non-auth error — verifies 404 is swallowed and CLI stays alive
• Updated test: tail-log - config — updated to not await runResult (loop never terminates); all 5 assertions still pass
• Existing test: tail-log - should retry 5 times and throw error — unchanged, MAX_RETRY_REACHED (has error.code) still re-thrown correctly
• Manual: run aio cloudmanager:tail-log before midnight UTC and confirm stream continues through rotation

🤖 Generated with Claude Code