iflib: update rs_pending based on descriptors consumed by isc_txd_encap

[gallatin]

Summary

The new watchdog code is triggering spurious watchdog resets on my out-of-tree NICs using ktls.

The issue is that rs_pending is updated using the descriptors for the current packet, based on what bus_dma produced.
However, NICs which support ktls offload may do extra DMAs (and consume extra segs) to derive crypto state when re-transmitting
TLS segments. This is the reason for allowing drivers to control ift_pad. When this happens, rs_pending may
be undercounted. This was harmless before, but is causing spurious watchdog resets now.

Fix this by using the actual segments consumed by the NICs isc_txd_encap.

Note this removes this historical overcount by one.

This patch has been stable on my test hardware (in combination with https://reviews.freebsd.org/D59204)

[kbowling]

I think this is a strict accounting improvement. e1000 manufactures some descriptors in a couple situations to work around hw errata, although I'm not sure that direct RS wb would've faulted.