Fix ch5 build hang: add network timeouts and reduce memory

[tyfolino]

Summary

The nightly build was hanging for 6+ hours (until GitHub cancelled it) due to missing network timeouts in Chapter 5, not memory.

• Root cause: urllib.request.urlopen() had no timeout — if AWS S3 was slow, the build would wait forever. s3fs.S3FileSystem also had no timeout configured. Ch3 already uses timeout=5 on its urlopen() call; this PR brings ch5 in line.
• Secondary bug: A 2-hour freshness check on the most recent MRMS file had no else branch, leaving data undefined if the file was stale → NameError in the plotting cell.
• Memory: Added data[::4, ::4] decimation after loading (~17x reduction, ~5 GB → ~300 MB) to keep hvplot fast. The interactive plot is preserved.

Changes to ch5_realtimeData.ipynb

• Add client_kwargs={'connect_timeout': 30, 'read_timeout': 60} to s3fs.S3FileSystem
• Add timeout=60 to urllib.request.urlopen()
• Remove the 2-hour freshness check (always fetch the most recent available file)
• Decimate loaded data: data = data[::4, ::4]
• Add a clear RuntimeError guard before the plotting cell

Test plan

• Confirm the build completes in <15 minutes
• Confirm the interactive hvplot is present in the built page

🤖 Generated with Claude Code

[github-actions]

👋 Thanks for opening this PR! The Cookbook will be automatically built with GitHub Actions. To see the status of your deployment, click below.
🔍 Git commit SHA: d9829af
✅ Deployment Preview URL: https://ProjectPythia.github.io/mrms-cookbook/_preview/51

[tyfolino]

Note for reviewers: If AWS S3 is temporarily slow at the time of a nightly build, the 60-second timeout on the download will cause the build to fail with a RuntimeError rather than hanging for 6 hours. This is an acceptable trade-off for now — a future improvement could add retry logic or fall back to cached outputs to handle transient AWS slowdowns more gracefully.