Data gap in PNGs suggests QC alert feature

[scottveirs]

Randomly checked the latest noise plots for Port Townsend this morning and noticed for the first time what looks like a data gap around 3am to 5:30am local (Pacific time):

I used my ts2mp3.sh script to pull the relevant ~6 hour data set from Port Townsend (epoch datetime stamp = 1717655423 ) -- which should cover 23:30 last night to 05:30 this morning. Viewed in Audacity, the concatenated .mp3 audio appears to be continuous (no obvious jumps in gain or discontinuities), so there appears to have been a data gap at after `3.5 hours of recording:

This suggests a possible new feature: this Github Action could be used to detect when one of the hydrophone nodes has failed to send audio data to the S3 bucket. It would be interesting to also experiment with unplugging the hydrophone to assess what the plots look like when there are still audio data segments coming in, but they have zero amplitude instead of a typical ocean signal.