* For everyone who gave me suggestions for my unexpectedly slow execution times issue, I wanted to follow up with the solution.

I turns out it was XIP--the Pico 2's flash is not able to keep up with the μC and running instructions directly from it introduces wait states--perhaps obvious in hindsight--but thank you James Munns for the XIP suggestion!  I'm not sure how long it would have taken me to suspect flash itself, but I _know_ it would have been a while!!  😅

My `cycle_count()` routine is both marked to be placed in SRAM by the linker and with the `#[inline(always)]` hint.  When it is being called from a function in flash, it will run from flash and will continue to be slow.  So once I ensured the caller was also running from SRAM, clock cycles between consecutive `cycle_count()` calls fell from 237 & 115 to 33 & 8 (debug & release, respectively).

Excerpt from `defmt` spew in release mode:
\`Instant::now().as\_ticks(): 75008231 INFO  Consecutive read 1 & 2: 75006709,75006717 = **Δ8cc**
└─ \<mod path> @ └─ \<invalid location: defmt frame-index: 7>:0

This is a great community--thanks to everyone for your help in figuring this out!