zephyr-cp: test_bsim_tx_power_low_rssi fails at an exact boundary (-39 < -39)

[mikeysklar]

Surfaced on PR #40 (zephyr rebase to zephyr_260803), CI run:
https://github.com/mikeysklar/circuitpython/actions/runs/31019513197/job/92352916509

Distinct from #42 — a different test, different failure shape (assertion, not a timeout).

FAILED tests/bsim/test_bsim_ble_advertising.py::test_bsim_tx_power_low_rssi - AssertionError: Expected lower RSSI with -20 dBm TX, got -39
assert -39 < -39

The test asserts a strict less-than against the low-TX-power RSSI reading, and the observed value lands exactly on the boundary rather than below it. Two live hypotheses, neither checked yet:

• A real regression from the rebase (something in the TX power path now reports/measures 1 dB high).
• The assertion itself is too strict for legitimate bsim RF-model noise, and this is the first rebase to actually land on the boundary.

Not yet root-caused. Full job run: tests / zephyr on PR #40, 4 failed, 108 passed, 10 skipped, 1 xpassed in 622.03s.