Preserve entity IDs in missing-type errors

[sapunyangkut]

Summary

• keep each entity's @id available while pick_type validates @type
• remove @id immediately after successful type selection, preserving normal construction behavior
• add regression coverage for both referenced data entities and contextual entities missing @type

Fixes #137.

Validation

• preserved evidence for this exact base and patch: targeted regression 2 passed; adjacent local read/profile/metadata tests 20 passed; narrow flake8 check passed
• git diff --check and reverse-apply check passed in the current worktree

The current system environment does not have the existing arcp dependency, so pytest cannot collect here; no dependency was installed. A broader preserved diagnostic is not claimed as fully green because pre-existing network-dependent read tests timed out.

AI assistance disclosure

This patch, tests, and pull request text were developed and checked with OpenAI Codex assistance. No independent human self-review is claimed; maintainer review is requested before merge.

[simleo]

Looks good, thank you. Before merging, I need you to update author and copyright info. See the discussion in #252 for instructions.

[elichad]

@simleo the account that created this PR is brand new (2 weeks old) and spent last week opening 80 PRs against many different Python repos. I've seen no evidence that there is any human involvement in its contributions.

It's up to you what you want to do with it - the PRs here look pretty harmless - but personally I don't look very favourably on PRs that aren't backed by a human who has reviewed the work before submitting it.