fix: UART parity mapping and digital-in reset before logic capture

[gabrielfrasantos]

Fixes two tickets, with one commit for each. They can be reviewed commit by commit.

1. LogicAnalyzer.arm() doesn't reset digital-in after the protocol decoders used it (f8cc4ae)

The UART, SPI and CAN protocol functions run on the device's digital-in resource and leave it configured for their own use. arm() sets only part of the digital-in state: acquisition mode, divider, sample format, buffer size and trigger. Everything else (sample mode, input order, other trigger details) was inherited from the protocol, so captures taken after a protocol had been used decoded wrongly.

Digilent's SDK reference says to call FDwfDigitalInReset before reusing digital-in after a protocol. arm() now makes that call first.

Test: test_logic_arm_resets_digital_in_left_configured_by_a_protocol configures and reads the UART, arms the logic analyzer, and asserts that FDwfDigitalInReset is the first digital-in call. It failed before the fix and passes after it.

2. UART parity mapping is swapped (0e2f644)

ProtocolUart.PARITY was {"none": 0, "odd": 1, "even": 2}. It is now {"none": 0, "even": 1, "odd": 2}.

The sources disagree on this, so it needs a careful review:

• Old mapping (1 = odd, 2 = even): the older WaveForms SDK manual, pydwf (which copies it) and Digilent's WaveForms-SDK-Getting-Started uart.py all use it. The bench's mapping came from there.
• New mapping (1 = even, 2 = odd): in this Digilent forum thread, a Digilent staff member calls the manual's order a documentation bug. The order they give is 0 none, 1 even, 2 odd, 3 mark, 4 space.
• The ticket's logic-analyzer captures are the deciding evidence. Please confirm on a real AD3 before merging.

I added a one-line comment on PARITY so the next reader doesn't "fix" it back from the manual.

Tests: test_uart_roundtrip now expects 1 for even. The new parametrized test test_uart_parity_uses_the_runtime_encoding covers none, even and odd.

Verification

• pytest: 127 passed, 1 skipped (the GUI test, because PySide6 isn't installed).
• ruff check and ruff format --check are clean.
• mypy reports 11 errors, all in gui/ because PySide6 isn't installed. main has the same errors.
• Not run on hardware.

🤖 Generated with Claude Code

https://claude.ai/code/session_01LKNpyfqJJLHgVZSg2kAePR

---

Generated by Claude Code