this is preparation for following PRs which add engine api tests (in
particular for reorg scenarios - relates to
#17025)

it extracts the MockCl from the shutter block building integration test
and generalises it a bit more so it is re-usable in more generic engine
api test scenarios (e.g. used in
#17105)

it also extracts the initialisation logic into a new EngineApiTester
which can also be re-used for different generic engine api test
scenarios (e.g. it can be used for the new "blockchain engine x test
format" - #16562)

this is preparation for following PRs which add engine api tests (in
particular for reorg scenarios - relates to
#17025)

it extracts the MockCl from the shutter block building integration test
and generalises it a bit more so it is re-usable in more generic engine
api test scenarios (e.g. used in
#17105)

it also extracts the initialisation logic into a new EngineApiTester
which can also be re-used for different generic engine api test
scenarios (e.g. it can be used for the new "blockchain engine x test
format" - #16562)

depends on #17103

relates to #17025 and
#17041

fixes repetetive error:
```
EROR[09-12|19:15:35.479] Failed to build PoS block                err="[1/3 MiningCreateBlock] wrong head block: 07ac2d701bd19780fc614cb658170f600ce8dd5dc9f4d7558226f43b315ed6fc (current) vs b2d30a3b6226d335a3786207b63ea029f59339551b06bbc528eeb563fbd50f96 (requested)"
```

we get into this state when (reproduced in the accompanying test):
1. we get a FCU for canonical block B at height H
2. and then get a NewPayload for a side chain block B' at block height H
3. and then get a FCU with head=B and payload attributes to start
building block at H+1

the reason for that is:
- when we process 2. we incorrectly unwind Erigon back to H-1 in order
to validate B' at H (validating a chain as part of NewPayload should not
affect the canonical chain, ie it should be "in-memory" or in disposable
tmp space)
- when we process 3. we get a quick payload status "VALID" for B at H
(because the fork choice stored in our DB still points to B at H) and we
then start building the new payload, however due to the incorrect unwind
our chaindata DB is at H-1 and so we get the `wrong head block` error
from the mining loop

the same error is seen in the issue with Prysm we had on fusaka-devnet-3
as described in #17025
(although this doesn't fully explain the wrong trie root we get when
trying to processes subsequent canonical payloads, so there might be
more unwind related bugs hidden elsewhere - tbd)

the same error fixes 1 out of 2 consistently failing tests (the
`/GetPayloadBodiesByRange \(Sidechain\) \(Paris\)` one) that we have in
the Hive ethereum/enginge withdrawals suite which we want to fix as per
#17041

Ive also seen this error happen sporadically in our Kurtosis tests
during reorgs making the tests flaky. This will stabilise those further
too.