fix(megasquirt): correct MIL channel data_type from boolean to bool

[SomethingNew71]

Corrects a one-character typo that was silently disabling the whole MegaSquirt adapter downstream.

The MIL channel declares data_type: boolean, but the adapter schema's enum is
["float", "int", "bool", "string", "enum"]. Across every adapter in this repo the value appears
307 times as float, 12 as int, 2 as bool, and exactly once as boolean — this line — so it
is a typo, not an intended variant.

It matters more than a single field suggests: a consumer deserializing data_type into a closed
enum rejects the entire adapter document, not just the offending channel. That is what
happened in UltraLog — megasquirt-tunerstudio.adapter.yaml failed to parse on every launch, and
all of its channel-name normalisation went with it. UltraLog has since been made tolerant of the
misspelling (ClassicMiniDIY/UltraLog#81), but the data should be correct at the source.

Other validator findings

While confirming this fix I ran every adapter and protocol against the schemas in schema/. The
repo does not currently pass its own validator, and there is no CI running it, which is why
this shipped. Nothing below is changed in this PR — flagging it separately:

Adapters — 6/9 valid

• megasquirt-tunerstudio — category: diagnostics is not in the schema's category enum, and
  file_format.header_row is required but absent
• haltech-nsp — header_row/data_start_row use -1 as a sentinel, but the schema sets
  minimum: 0; timestamp_column/timestamp_unit are null where a string is required
• emerald-lg — channels carry internal_id, which the schema disallows

Protocols — 4/9 valid

• ecumaster-emu-broadcast — configurable_base_id, default_base_id
• maxxecu-default — base_address, base_address_configurable
• emtron-broadcast — byte_order at protocol level
• megasquirt-broadcast — comment
• rusefi-broadcast — offset on messages

Most of these look like the schema lagging the data rather than bad data — diagnostics is a
real category (UltraLog's own type already supports it), -1 is a meaningful "auto-detect"
sentinel, and the extra protocol keys look deliberate. So the fix is probably to widen the schemas
rather than strip the fields, but that is a judgement call per field and belongs in its own PR.

Worth adding scripts/validate-all.sh to CI once the schemas and data agree, so the next typo is
caught before it reaches consumers.

[Copilot]

Pull request overview

This PR fixes a schema-invalid data_type value in the MegaSquirt TunerStudio adapter so downstream consumers that deserialize data_type as a closed enum don’t reject the entire adapter document.

Changes:

• Updates the MIL channel data_type from boolean to the schema-valid bool.

---

💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.