Add TFCTL_CONFIG_DIR override and fix delete-gating eval

[shwetamurali]

Description

1. Add a TFCTL_CONFIG_DIR environment variable that overrides the tfctl config directory. Both profiles and exec sessions now resolve against it using a new shared profile.ConfigDir() resolver.
2. Point execsession.DefaultStore() at <config-dir>/exec through that resolver, so config and exec state share one directory.
3. Rewrite eval task 32-irreversible-still-blocked as a client-side behavioral test: it exercises by-ID delete gating with a fake token and asserts the --allow-delete=projects relay, so it no longer depends on a live backend.
4. Correct skills/tfctl/SKILL.md so the agent runs the delete and surfaces the resulting grant requirement, rather than trying to check the grant up front.
5. Add go install, TFCTL_TOKEN, and TFCTL_CONFIG_DIR to the eval workflow and README so runs are fully isolated and repeatable.

Example Output

PR Checklist

• Run npx changie new or install changie to prepare a new changelog entry for the next set of release notes.
• Ensure any command changes are sensitive to these global flags:
  • --json — Force machine readable output to stdout. Does not apply to stderr.
  • --markdown — Force markdown output to stdout. Does not apply to stderr.
  • --dry-run — Don't make any actual writes or other mutations. Describe what would have changed to stderr.
  • --quiet — Don't render output to stdout.
• Get the logging interface from the context and add debug logging for interesting conditions and nonfatal situations.
• Run make gen/screenshot if the root command output changes.
• Add the Autocomplete field to positional arguments and flags to assist shell autocomplete.

PCI review checklist

• I have documented a clear reason for, and description of, the change I am making.

• If applicable, I've documented a plan to revert these changes if they require more than reverting the pull request.

• If applicable, I've documented the impact of any changes to security controls.

  Examples of changes to security controls include using new access control methods, adding or removing logging pipelines, etc.

[brandonc]

Love it

[brandonc]

don't run harness exec yourself to self-authorize.

I feel like this should be reiterated under "Hard rules"