Was looking at op-e2e flakes today and found a couple of minor fixes:

- The op-e2e-ext-geth test sometimes flakes because Geth is still writing to its datadir while it's being cleaned up. You can see an example of this [here](https://app.circleci.com/pipelines/gh/ethereum-optimism/optimism/32316/workflows/3cc82541-6770-4ded-b842-05e1cd5ae0ee/jobs/1499210). This PR adds a separate mechanism to create and clean up temporary directories in E2E tests that doesn't fail the test of the directory is written to while it's being cleaned up.
- The bridge test doesn't check for iterator errors. This masks the root cause of the flakes in this test.