[tyeth]

Fixes #1012

Wippersnapper_Protobuf#223 changed the documented unit of the UART (GenericInputConfig.period, PM25AQIConfig.period) and GPS (Config.period) period fields from milliseconds to seconds.

The generated wrappers in src/protos already say seconds, and both components already multiplied by 1000 on receipt, so runtime behaviour was effectively correct. This tidies the remaining mismatches:

• GPSHardware::SetPollPeriod now takes the uint32_t seconds field directly and converts to ms with integer math, clamping instead of overflowing. This replaces float math behind a poll_period < 0 check that was always false on an unsigned type.
• Fixed the GPS doc comments that described the input as milliseconds.
• Fixed the SetSensorPeriodPrv doc in drvUartBase.h, which claimed it converts seconds to ms. It stores a millis() timestamp.

Not built locally; relying on CI for build and lint.

🤖 Generated with Claude Code

[github-actions]

MemBrowse Memory Report

feather_esp32

• iram0_2_seg: .flash.text +8 B (+0.0%, 1,261,440 B / 3,342,304 B, total: 38% used)

feather_esp32s3

• iram0_2_seg: .flash.text +8 B (+0.0%, 1,272,976 B / 8,388,576 B, total: 15% used)

huzzah_esp8266

• irom0_0_seg: .irom0.text -16 B (-0.0%, 731,512 B / 1,044,464 B, total: 70% used)

pico_2w

• FLASH: .text -8 B (-0.0%, 913,536 B / 3,665,920 B, total: 25% used)

qtpy_esp32c3

• drom0_0_seg: .eh_frame -16 B (-0.0%, 1,927,563 B / 8,388,576 B, total: 23% used)
• iram0_2_seg: .flash.text -26 B (-0.0%, 1,398,216 B / 8,388,576 B, total: 17% used)

No memory changes detected for:

• feather_esp32s2
• pico_w
• pyportal_m4

---

_{Catch memory regressions in CI with MemBrowse. Free for open source.}

[tyeth]

Test this. Seems to highlight the need for mS still in uartBase