Fix error when starting work on a repo that was private but is now public

[zanerock]

Overview

The following resulted in an error:

1. Create private repo.
2. Start, submit, and clean work. (May not be necessary to reproduce?)
3. Switch repo to public via git interface.
4. Start work.

Got error:

Failed to execute 'cd '/Users/zane/.liq/playground/liquid-labs/npm-toolkit' &&
git checkout -b work-liquid-labs/npm-toolkit/3 && git push --set-upstream
workspace work-liquid-labs/npm-toolkit/3'; stderr: Switched to a new branch
'work-liquid-labs/npm-toolkit/3'
ERROR: Repository not found.
fatal: Could not read from remote repository.

Please make sure you have the correct access rights
and the repository exists.

Stacktrace:

at Module.tryExec (/Users/zane/.liq/playground/liquid-labs/liq-work/node_modules/@liquid-labs/shell-toolkit/dist/shell-toolkit.js:27:39)
    at /Users/zane/.liq/playground/liquid-labs/liq-work/dist/liq-work.js:310:303
    at h (/Users/zane/.liq/playground/liquid-labs/liq-work/dist/liq-work.js:18:91)
    at Generator.<anonymous> (/Users/zane/.liq/playground/liquid-labs/liq-work/dist/liq-work.js:38:7)
    at Generator.throw (/Users/zane/.liq/playground/liquid-labs/liq-work/dist/liq-work.js:25:93)
    at g (/Users/zane/.liq/playground/liquid-labs/liq-work/dist/liq-work.js:11:580)
    at i (/Users/zane/.liq/playground/liquid-labs/liq-work/dist/liq-work.js:13:51)

It was observed that a 'workspace' remote did exist (checked after the first 'issue start'), but the URL was messed up: git@github.com:undefined/npm-toolkit.git. The 'undefined' bit should have referred to the user's github account.

Setting the branch back to main, deleting the workbranch, and updating the remote URL fixed the issue. There was a fork created (checked after first 'start issue') at some point.

Related issue 1: when re-running the command, we got a different error because we were now on the work branch. We need to verify the remote looks good before switching branches or set self back to main after error.

Related issue 2: attempting to 'pause' the work resulted in a complaint that there was no unit of work with that name, so the branch was being created before the unit of work. It should probably go the other way and the unit of work should be created early in the process after validations have been performed but before other work.

Deliverables

• Do this first. Update flow to verify remote exists and is accessable before switching to the work branch OR implement cleanup in the face of an error. (This may or may not be entangled with the primary issue.)
• Determine cause of error.
• Fix issue.