[jtag_dmi_reg] Correctly handle a confusing situation where TCK is much faster than the main clock

[rswarbrick]

This fixes issue #21517. The first two three commits have no functional change, but make the code a bit easier to read.

Following that is a change to the frontdoor sequence. This tweak ensures that we send an extra JTAG transaction just after doing a DMI request. Doing so ensures that we get truthful tracking of DmiOpInProgress afterwards.

Finally, there's a change to the monitor to avoid it believing that the DUT has suddenly become busy without being asked to do an operation (which is very confusing to debug!)