Regression in master merged with PR #7022: WDT resets

[dok-net]

Details in #7022 and #7121.
PR #7022 works without lock-ups or reboots up to and including master commit 2c435b1. So commit c61b70d introduces some change that breaks the timing and causes continual WDT resets (1 per minute or so) in the MVCE I've supplied in #7121.
Without PR #7022, I have not been able to reproduce this malfunction, but that shouldn't make it any less significant.

[dok-net]

@d-a-v The connect to the PR #7022 was mentioned in #7121 ("waveform" intended to mean the waveform PR, made clear by the immediate reference to PR #7022 in the same breath). Sorry to have caused any distress over this.

[dok-net]

To be clear, OTA is included in the MCVE because it's in the actual complete code that I am working with. OTA is not in use when the WDT triggers, just the requisite call in loop() is there.
If OTA is removed from the sources, the WDT does not trigger. If only #7036 is reverted, the WDT still triggers, but is observed less frequently. Reverting #7121 mitigates the WDT issue completely.

[dok-net]

WDT resets were fixed by #7121, then finally merged via #7131