feat: add GCP monitoring module and alert policies for DE-04-T2

[IsurangiGuniyangodage]

Why

DE-04-T2 requires monitoring and alerting for scraper/ETL pipeline failures, following the GCP-native approach recommended in DE-04-T1. Pipeline failures currently go unnoticed unless manually checked.

What changed

• Added discount-mate-infra/modules/monitoring: four log-based metrics (scraper_rows, scraper_run_failed, scraper_http_errors, scraper_block) and three alert policies (job failed, zero/near-zero rows with missing-data-as-breach, HTTP error-rate spike)
• Slack (#alerts-scrapers) and email notification channels, per the implementation plan's redundancy recommendation
• Wired the module into environments/prod (enabled monitoring.googleapis.com, added monitoring_alert_email variable)
• terraform fmt and terraform validate both pass
• Instrumented all four scrapers (IGA, Woolworths, ALDI, Coles) with structured JSON logging: each now emits a scrape_run_finished summary (rows scraped, duration, HTTP status breakdown, retries), a scrape_block event on detected soft-blocks/403s, and a scrape_run_failed event on unhandled exceptions — the data the log-based metrics above depend on
• All four scraper files verified to compile cleanly under Python 3.12 (the pipeline's required version)

Notes

Infrastructure and scraper-side logging are both complete and internally consistent — the alert policies now reference log fields the scrapers actually emit.

Verified locally: set up a proper Python 3.12 environment for the pipeline using uv sync (per its pyproject.toml/uv.lock), then confirmed all three logging functions (emit_scrape_summary, emit_scrape_block, emit_scrape_run_failed) produce valid JSON with the exact fields the Terraform log-based metrics expect (rows_scraped, event, scraper, http_403, etc.) for both success and failure cases.

terraform apply against live infra and fault-injection testing (confirming each alert policy fires correctly end-to-end) will be handled by Ben Van. Marking as ready for review.

Update: Refactored scraper run-stats tracking into a new RunStats model attached to RunContext, per Ben's review feedback — applied consistently across all four scrapers (IGA, Woolworths, ALDI, Coles), removing the extra try/except layers.

[IsurangiGuniyangodage]

@callmesoffie1811 and @anhvdq Please have a look!

[IsurangiGuniyangodage]

I updated the PR. Please have a look @callmesoffie1811 and @anhvdq

[anhvdq]

Thanks for your works on this PR
I put a comment on how we collect the stats of the jobs, it would be better to implement in the RunContext instead of current try-catch block

Besides, the terraform code seems good to go

[anhvdq]

The changes looks good on the IGA and WW scrapers but seems like Aldi and Coles were not updated yet
Could you take a look and fix them?
Then all good to go

[anhvdq]

I have fixed some issues with the format and try-catch for better practices
All good now