docs(readme): rewrite positioning, add evaluation architecture, and motivation

[theonly1me]

What changed

• Rewrite README.md to lead with the falsifiable behavioral claim and the shadowclone eval command.
• Define architectural scope explicitly as a memory and alignment compiler rather than an agent runtime.
• Document ground-truth privacy guarantees including sliced secret redaction, Shannon entropy filtering, and categorical tool result exclusions.
• Document enterprise governance ceilings via root-owned managed configuration files.
• Author open-source motivation and design philosophy in docs/motivation.md.
• Add replay evaluation architecture doc in docs/architecture/09-evaluation.md and update architecture index.

Why

Leading with falsifiable behavioral metrics gives engineers a concrete reason to trust shadowclone over uncalibrated memory tools. Clarifying that shadowclone compiles memory for existing CLIs rather than replacing them prevents category confusion. Documenting privacy architectures and managed policies gives security reviewers verified claims they can audit directly against the source.

How to verify

bun run check

Data handling

• Any new capture source has an opt-in flag and a README entry in this PR
• Everything that reaches the network passes redactSecrets
• No raw capture in a log line, an error message, or a test fixture
• A test exercises the real entry point, not just the redaction function

[theonly1me]

We already have claude code auto-memory for this, I think what we're solving isn't solved by that.