This change reduces the default release artifact size by enabling standard Rust release-profile optimizations at the workspace root.

Before this change, the release binary was large even before any feature-level decomposition because the build was not stripping symbols and was not using a link-time optimization profile that favored a smaller final artifact. That meant users were paying a significant size cost regardless of which channels they actually configured at runtime.

The root cause is straightforward: the repository did not define an explicit [profile.release] section in Cargo.toml, so the build was using the default release profile without strip, without LTO tuning, and without constrained codegen unit settings. For a binary with a large dependency graph, that leaves a meaningful amount of size on the table.

This PR adds:

• strip = true
• lto = "thin"
• codegen-units = 1

The practical outcome from a clean local cargo build --release on Apple Silicon was:

• target/release/microclaw: 91M -> 57M
• release tarball: 33M -> 24M

This PR intentionally does not change feature topology or split channel integrations yet. That follow-up work is tracked in #309, which covers making channel-specific integrations optional so that unused channels do not contribute to the default artifact size.

Validation for this PR was limited to a clean release rebuild and artifact measurement:

• cargo clean && cargo build --release
• repackaged target/release/microclaw into a release-style .tar.gz and compared sizes