[adamshapiro0]

New Features

• Add InputDataWrapperMessage.count_wrapped_data() to total up wrapped packets by type straight from a file, without deserializing each message

Changes

• Generate plots roughly 1.6x faster and with less than half the peak memory on a multi-hour log
  • Skip building per-message debug/trace strings when they won't be printed
  • Reduce MixedLogReader progress check call frequency
  • Limit unnecessary, expensive enum __call__() operations with new lookup_value() function
  • Improved p1_display summary reset event table generation speed
  • Reuse satellite and signal ID instances across epochs
  • Release decoded GNSSSignals messages when not needed
  • Build constant trace values as numpy arrays rather than lists
  • Add orjson package for faster Plotly JSON encoding
• Reduced plot HTML file sizes (up to 40% or more for long logs)
  • Share a single customdata across multiple traces in a plot
  • Write one shared plotly.min.js alongside the figures instead of embedding a copy in each one, so an individual plot file now needs the rest of the output directory to render
  • Shrink the generated HTML by emitting each figure's shared hover time data once rather than once per trace
  • Write the UTC time axis at millisecond resolution, which is all a browser can display
  • Encode UTC X-axis values as base64-encoded float instead of date strings (the Plotly default)
  • Share identical X-axis arrays between Plotly traces in the generated HTML output

Fixes

• Fix EventNotificationMessage.to_numpy() raising OverflowError on numpy 2 when a log event carries a negative severity
• Fix logger.trace() raising TypeError when passed printf-style arguments
• Fall back to any source ID when plotting map if no IDs found in pose source range
• Pin Plotly version <7 to keep the generated map loading from disk (file:// no longer works with Plotly 7)

[bhers4]

Looks good to me!