feat: add CLI control for LLM call details

[regnull]

Summary

Add explicit backend CLI flags for detailed LLM call activity data and make Electron opt out by default.

Problem

Electron currently starts the backend with detailed per-call usage data enabled unconditionally, while there was no documented command-line control.

Solution

• Add backend.openbot.cli with --include-llm-call-details and --exclude-llm-call-details.
• Preserve browser and make run behavior by keeping the backend setting enabled by default.
• Start Electron with details excluded unless --include-llm-call-details is passed.
• Forward the setting through the packaged backend launcher and document the opt-in.

Files changed

File	Change
backend/openbot/cli.py	New CLI entry point and flag parsing.
backend/openbot/config.py	Adds the authoritative setting.
backend/openbot/runtime/runner.py	Suppresses usage detail payloads when disabled.
scripts/electron-dev.sh	Adds Electron opt-in forwarding and default exclusion.
scripts/electron-backend.sh	Forwards packaged startup setting.
Makefile, README.md	Routes backend launch through CLI and documents behavior.

Validation

• Python compilation passed.
• Shell syntax checks passed.
• git diff --check passed.

Full repository lint/test/build and CI checks remain to be run.

[regnull]

Changes required.

1. BLOCKER — PR #150 — README.md:67 (and Makefile:17): the documented opt-in command make electron -- --include-llm-call-details does not forward the flag to scripts/electron-dev.sh. GNU make treats --include-llm-call-details as an additional goal, so this command starts the target without the argument and then fails with No rule to make target '--include-llm-call-details'. The script's argument handling only works when invoked directly. Please make the documented make electron form actually forward the opt-in (or change the documentation to a working invocation and add a regression test for that invocation); retain the default exclusion behavior.

— Reviewer - @Reviewer

[regnull]

Ready to merge.

Re-reviewed PR #150 at head 80e2c00. The Make forwarding fix is present in the actual diff, and I verified both commands: make -n electron -- --include-llm-call-details emits ./scripts/electron-dev.sh --include-llm-call-details, while plain make -n electron emits no opt-in. The launcher defaults to exclusion and the focused regression coverage is included.

No blocking findings or additional nits.

— Reviewer - @Reviewer