WIP DO NOT MERGE Fix: save_map_binary wrote scenario_id prefix that C never reads

[eugenevinitsky]

Summary

save_map_binary wrote 16 bytes of scenario_id before sdc_track_index, but load_map_binary in C starts reading at sdc_track_index with no scenario_id. This misaligned all subsequent reads, causing segfaults when loading converted maps.

Also fixes Town10HD conversion failure (scenario_id was an int, not str).

Test plan

• Convert carla_data maps with process_all_maps
• Verify converted maps load without segfault
• Verify existing carla_2D/carla maps still work

[Copilot]

Pull request overview

Fixes a binary format mismatch between Python map conversion (save_map_binary) and the C loader (load_map_binary) that was misaligning reads and causing crashes when loading converted maps.

Changes:

• Removed writing a 16-byte scenario_id string prefix in save_map_binary so the file starts with sdc_track_index, matching the C reader.
• Avoided conversion-time failures when scenario_id isn’t a string by eliminating the .encode() path entirely (since the prefix is no longer written).

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

This change alters the on-disk binary format (removing the 16-byte prefix). Please add a small regression test that writes a minimal map via save_map_binary and asserts the first bytes are sdc_track_index (i.e., no 16-byte offset) so this misalignment doesn’t get reintroduced later.

[Copilot]

The note here is easy to misread as “scenario_id isn’t part of the binary at all”, but load_map_binary does read an int scenario_id per-entity. Consider rewording to explicitly say “the file must start with sdc_track_index (no 16-byte string prefix)” and/or reference pufferlib/ocean/drive/drive.h:1060 to prevent confusion.

Suggested change

	# Note: C load_map_binary does NOT read a scenario_id prefix.
	# Do not write one here or the binary will be misaligned.
	# Binary layout note:
	# The file must start with sdc_track_index (an int32), with NO 16-byte
	# scenario_id string prefix. The corresponding C reader
	# (load_map_binary in pufferlib/ocean/drive/drive.h:1060) reads
	# scenario_id as an int per-entity, not as a file-level prefix. Writing
	# any extra prefix here will misalign all subsequent reads.