Running them sequentially actually turns out to be faster. In addition it
turns out to be very hard to split the tests properly. `go list ./...` which
was the previous approach splits them by package, but we need to split them
by test name. Using `circleci tests glob "**/*.go"` did not work either
becaue it was combining files from different packages.

TestConfirmationDepth was consistently failing on the HTTP versions of the
op-e2e test. It was failing at the last line that checked the verifier
head block timestamp against the L1 head timestamp. I have changed it to
compare the verifier L1 Origin number against the L1 head number.

feat(ctb): add MigrationSystemDictator checks

Adds checks for the MigrationSystemDictator steps.

Version Packages

Develop -> Master PR

op-e2e: Add migration to op-e2e

hardhat-node: trigger release

op-e2e: action test op-geth restart

Trigger a release

batch-submitter: update to go-ethereum v1.10.26

The `hardhat-node` docker image couldn't find the
`Dockerfile` when building. This updates the path to
the dockerfile.

Updates the batch submitter to the latest go-ethereum
version

Develop -> Master

Version Packages

Develop -> Master PR

fixes a few bugs and better error handling for docker-tag job

fix missing directory in docker-tag job

ENG-2990 updates ci to tag docker images with sha, branch, release tags

Standardize & Fix Lints

fix: move MintableERC721 interface

op-e2e: Increase sequencer drift in sequencer conf depth test

The test was flaking on the issue that l2 verifier head + conf depth
was ahead of the sequencer head. I noticed that we were occasionally
having to force advancing the L1 Origin because of the high conf depth
relative to the sequencer depth.

ci: Split go-bedrock tests & use circle CI parallelism for op-e2e

Adding OP_E2E_DISABLE_PARALLEL=true will stop the tests from using go's
built in parallel helper. By default the tests will use go's parallel helper.