[DON'T MERGE] Incident intervals prototype

[zilm13]

Implementation of incident intervals in Teku

• Dirty hack is made to use incident intervals only after client restart, because I didn't find a better way to use different flags/config in Kurtosis only after restart in the middle.
• 2 changes are not yet added to the spec and should be clarified
  • exception when trying to produce Attestation in incident interval in getGenericAttestationData
  • failed broadcast validation in BlockManager for blocks inside incident interval, so local block produced during incident interval is not published (otherwise it will be published even with failed block import)

recover.yaml for Kurtosis

participants:
  - el_type: geth
    cl_type: teku
    cl_extra_params: [
        --validators-keystore-locking-enabled=false,
	--Xincident-intervals=1-2
    ]
    cl_image: consensys/teku:develop
    cl_min_cpu: 2000
    cl_min_mem: 2000
    count: 5
network_params:
  electra_fork_epoch: 1
global_log_level: info
additional_services:
  - dora
  - spamoor_blob
  - prometheus_grafana

Steps to reproduce expected scenario:

# build Teku and save in local Docker repository
% ./gradlew distDocker 
# start Kurtosis with the provided config
% ./kurtosis run --enclave my-testnet github.com/ethpandaops/ethereum-package --args-file recover.yaml

On the start of epoch 1 stop 2-5 nodes and start them again (so incident intervals will be applied after restart):

% kurtosis service stop my-testnet cl-2-teku-geth                                                  
% kurtosis service stop my-testnet cl-3-teku-geth
% kurtosis service stop my-testnet cl-4-teku-geth
% kurtosis service stop my-testnet cl-5-teku-geth


% kurtosis service start my-testnet cl-2-teku-geth
% kurtosis service start my-testnet cl-3-teku-geth
% kurtosis service start my-testnet cl-4-teku-geth
% kurtosis service start my-testnet cl-5-teku-geth

[zilm13]

Expected behavior:

chain created by node-1 during epochs 1-2 is orphaned

chain derived on the side from epoch 3 is finalized

There are some issues with prototype so we don't see 100% at this moment, but I will not focus on this issue at the current stage (need to abandon already imported blocks during incident intervals probably)

[rolfyone]

hrm so thats interesting, im not sure any epochs were completely empty...

[zilm13]

it's expected that it must fork so all the slots inside incident intervals are completely empty