For what it's worth, here's my latest results with Glo #4. No reading, only turning on to record battery state.
Device was left in the (no magnet) cover:
• March 20 – 4:00 AM - 100%
• March 24 – 3:00 PM Dead
Removed device from cover:
• March 28 – 11:20 PM 100%
• March 29 – 11:24 AM 100%
• March 29 – 11:45 PM 100%
• March 30 – 12:23 PM 100% (changed books, and flipped +/- 20 pages)
• March 31 – 11:08 AM 93%
• April 1 – 4:00 AM 93%
• April 1 – 7:36 PM 93%
• April 2 – 7:28 PM 93%
For the out-of-cover tests, I put the device to sleep with no movement in front of cover for approx. 5 seconds before putting device down.
Thanks to everyone else in this thread for their tests, comments and conclusions.

I hope a fix isn't too far off.