op-program,op-service: dedup BlockInfo header wrapper, always cache hash, avoid recomputing hash (#13681)

* contracts: Add gas input to precompile pre-images (#186)

Also update the cannon evm tests to use the new precompile preimage scheme.

---------
Co-authored-by: Adrian Sutton <adrian@oplabs.co>

* op-challenger: Support uploading data in new format. (#188)

* op-program: Add required gas to precompile oracle key (#176)

* op-challenger: Support multiple versions of the preimage oracle contract

---------
Co-authored-by: Adrian Sutton <adrian@oplabs.co>

---------
Co-authored-by: inphi <mlaw2501@gmail.com>

---------
Co-authored-by: Adrian Sutton <adrian@oplabs.co>

* op-program: Generic precompile oracle

The generic precompile oracle replaces the point evaluation precompile
oracle. The new oracle can be used to retrieve the return data and
status of any precompile, including bn256Pairing and ecrecover.

* op-challenger: Use generic precompile oracle

Replace the KZG point evaluation oracle with a generic precompile oracle

* op-program: Simplify required gas logic.
Use FromHex instead of Hex2Bytes

* op-program: Set beacon URL when capturing and verifying sepolia chain data

* op-program: Actually use the beacon URL

* op-program: Use default l1-rpckind if not set rather than overriding to debug_geth

* Use freshly generated sepolia test data

* Use sepolia compatibility data

* Fix spelling

* update PreimageOracle.sol snapshot

---------
Co-authored-by: Adrian Sutton <adrian@oplabs.co>

* FP: add KZG point-evaluation preimage-oracle type, to process L2 oracle call with

* op-program: Integrate kzg precompile oracle

* fix PreimageOracle.sol

* store kzg precompile input preimage

* update bindings and go.mod

* fix go.mod

* resolve TODOs

* s/Hex2Bytes/FromHex

* fix typo

* ctb: set kzg preimage length to 1

* op-challenger: Load kzg point evaluations to PreimageOracle (#9497)

* op-challenger: load KZG point evaluation into oracle

Also: increase max game depth for devnet.
This is a temporary solution to avoid flakes in e2e tests that use the kzg precompile.
The current execution trace max step of 2**31 steps isn't sufficient.

---------
Co-authored-by: inphi <mlaw2501@gmail.com>
Co-authored-by: refcell <abigger87@gmail.com>

* Add blob preimage type

* 4844 blob fetcher

key type

* use params package constants

* Add l1 ref timestamp to commitment hint

* single preimage hint

add log

* Stub oracle

* Pass in blobs fetcher

* Encode hex in string hint route

* Panic if blob reconstruction was faulty

* @refcell review

* rebase

The output root replaces the L2 head hash at boot as the agreed upon point to
run the fault proof program. The L2 head is still necessary for
derivation but is redundant as it's contained in the l2 output.