Build nRF52840 DK fails: variant nRF52DK absent in installed Adafruit framework

[zackees]

Symptom

Build nRF52840 DK fails on every main push with:

fatal error: variant.h: No such file or directory
cores/nRF5/Uart.h:27:10
   27 | #include "variant.h"

Run: https://github.com/FastLED/fbuild/actions/runs/26694160451

Root cause

fbuild ships one nRF52 framework — Adafruit's framework-arduinoadafruitnrf52 (hardcoded URL in crates/fbuild-packages/src/library/nrf52_core.rs).

The nrf52840_dk board JSON declares build.variant = "nRF52DK" (matching PIO upstream — the board-validator agrees). But the Adafruit framework doesn't have a variants/nRF52DK/ directory. It has variants/pca10056/ instead — PCA10056 is Nordic's product code for the nRF52840-DK, so it's the right hardware. Verified via GitHub API: GET /repos/adafruit/Adafruit_nRF52_Arduino/contents/variants?ref=1.6.1 lists pca10056, no nRF52DK.

variants/nRF52DK/ is in the other nRF52 Arduino framework: sandeepmistry's arduino-nRF5 (the one PIO actually selects for this board). fbuild doesn't ship that one.

Result: cores/nRF5/Uart.h does #include "variant.h", gcc's include search walks the (nonexistent) variants/nRF52DK/ directory, finds nothing, link breaks.

Fix options

Option	Pros	Cons
A. Per-framework variant aliases: add a small nRF52DK → pca10056 map in SamdCores::get_variant_dir's nrf52 counterpart. Falls back when the literal name is missing.	Surgical; preserves PIO-matching board JSON; doesn't pull in a second framework download.	Each new "fbuild claims it but Adafruit framework has a different name for it" board needs an entry.
B. Install sandeepmistry's framework alongside Adafruit's; pick based on board.	True "supports the full nRF52 board catalog".	New download, new resolver, ~3x more code; the sandeepmistry framework is in maintenance mode and lags.
C. Change nrf52840_dk board JSON to variant = "pca10056".	One-line fix.	Breaks board-validator (diverges from PIO upstream); makes the board JSON lie about its PIO compatibility.
D. Remove nrf52840_dk from the test matrix and bundled boards.	Eliminates the broken claim.	Loses a board fbuild was supposedly supporting.

Recommended: A. Same shape as the #319 fix for SAMD's get_core_dir (Adafruit-branded build.core = "adafruit" falling back to cores/arduino/). Both bugs are the same pattern: PIO upstream names a thing one way, the installed framework names it differently, fbuild's lookup didn't paper over the difference.

Test plan

• Unit test get_variant_dir fallback: given a synthetic framework with only variants/pca10056/, get_variant_dir("nRF52DK") should resolve to .../variants/pca10056/.
• Integration: Build nRF52840 DK workflow turns green on next main push.
• Sanity-check the other nRF52 workflows (Adafruit Feather NRF52840 Sense, etc.) continue to pass (their board JSONs use Adafruit-native variant names so they should be unaffected by the fallback).

Refs #319 (same shape, SAMD platform).