Add proper support for blobs in JSON-RPC methods

[fvictorio]

See https://github.com/NomicFoundation/hardhat/issues/4840

This needs to be defined. Some initial questions:

• Do we have a separate mempool for blob transactions, as clients have? How are they prioritized when mining?
• Given that that seems hard to get right, should we start with an implementation that only works with auto-mining?
• Do we "realistically" drop blob data, or do we just keep it, since it's a development network?
• Do we need to add any non-standard methods?