rp2: Bump pico-sdk to v2.3.1 - #19683

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.59%. Comparing base (0fd6c57) to head (3a07249).
⚠️ Report is 71 commits behind head on master.

@@           Coverage Diff           @@
##           master   #19683   +/-   ##
=======================================
  Coverage   98.58%   98.59%           
=======================================
  Files         182      182           
  Lines       23322    23335   +13     
  Branches        5        5           
=======================================
+ Hits        22993    23006   +13     
  Misses        328      328           
  Partials        1        1           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

Code size report:

Reference:  rp2: Page-align the C heap ceiling. [11094ea]
Comparison: rp2: Document libc/mem-in-RAM divergence from pico-sdk 2.3.1. [merge of 3a07249]
  mpy-cross:    +0 +0.000% 
   bare-arm:    +0 +0.000% 
minimal x86:    +0 +0.000% 
   unix x64:    +0 +0.000% standard
      stm32:    +0 +0.000% PYBV10
      esp32:    +0 +0.000% ESP32_GENERIC
     mimxrt:    +0 +0.000% TEENSY40
        rp2:   -68 -0.007% RPI_PICO_W[incl -4(bss)]
       samd:    +0 +0.000% ADAFRUIT_ITSYBITSY_M4_EXPRESS
  qemu rv32:    +0 +0.000% VIRT_RV32

Will have to poke that rp2 size difference and see if any of it is linker script wobbles versus the code refactor.

Octoprobe PR report

• Duration: 33min, 13991 tests run, 143ms/test
• Tested ports: rp2
• Summary Report , Logdirectory

Regarding the above Octoprobe results (the second edit to the comment):

• the mpremote test failures seem to be specific to the Octoprobe setup, I cannot reproduce them locally (tested on RPI_PICO2 and RPI_PICO2-RISCV)
• the BLE failures are a flaky test that failed 1/3 times and can be considered a pass (would also probably be flaky on pico-sdk 2.3.0)
• the net_hosted/ssl_verify_callback.py failure is a true failure that also happens on master and is not related to this PR

I've dropped an extra linker .incl which was in this PR but was not only doing basically nothing but would have exploded SRAM usage on RISC-V with a C++ module linked.

The *libc.a divergence (which still works) never matched anything on Arm builds; all the newlib symbols resolve from libg.a. For RISC-V it moves the following into SRAM (and always has afaik):

If we were to fix it for Arm, I'd use roughly ~1.2 KB SRAM which is a per-board decision a downstream vendor (like us) might want to make (at least before we ran out of SRAM), but not a suitable cost for RP2 in general.

The perf boost isn't all that great anyway (that wild misc_mandel swing is a particularly egregious case of flash placement lottery):