Repair the latest known queue index on startup to prevent
skipped deposits

This will ensure the `builder` is built first so that other
services that depend on the `builder` do not pull a remote
`builder` image.

Adds the http header 'User-Agent' to the creation
of ethers providers across the services to allow
for more visibility into what services are sending
requests.

Also delete dead imports since they were breaking
during linting.

Move the instantiation of the providers to earlier
in the script so that they can be used to check the
chain id before doing any slow disk reads of giant json
files.

Just the the `ethProvider` to deduplicate usage of a provider.
They both are connected to the same thing, so this removes config.

The `GenesisJsonProvider` implements the `ethers.Provider`
interface and is constructed with a geth genesis file, either
as an object or as a file to be read from disk. It implements
a subset of the RPC methods that use the genesis file
as the backing storage. It includes tests for its correctness.
Not all methods are implemented, just the ones for the regenesis
testing.

This PR also moves the tests around in the `core-utils` package
as some of the tests were being skipped. The `tests` directory is
flattened, having so many subdirectories was not needed. The
`package.json` test script is updated to ensure that all tests
are run.

Also add some deps that are required for the `GenesisJsonProvider`.

Ensure that test coverage output is not linted, or formatted,
and is removed by the clean script.

which fixes https://github.com/sc-forks/solidity-coverage/issues/652

Can bring up the system without needing to build locally
using the following command:

```
$ docker-compose -f docker-compose-nobuild.yml up
```

Simply clean up the `time.Ticker`s using `defer t.Close()`.
Not having this code should not cause any problems, but
it is best practice to close the tickers as not doing so
can result in memory leaks in certain cases.

Prevent the error being returned in consensus for when
the user does not have enough balance. The policy level
check *must* filter out transactions that do not have
enough gas to pay for their execution. The policy
level check happens in the `SyncService`. This change
gracefully handles the race condition where a user transaction
passes the policy level check as a `gas-oracle` transaction
is executing and updating the gas prices. If the gas price
goes up and the user transaction no longer can afford execution,
then it will return an error from consensus.

Adds the config option `DATA_TRANSPORT_LAYER_STARTING_L1_BLOCK_NUMBER`
or `--starting-l1-blocknumber` that allows for the user to specify
the L1 blocknumber to begin syncing from. This prevents users from
syncing old batches from a previous regenesis.