Omni-node: Move timestamps closer to now

[skunert]

In omni-node dev-mode the timestamps where showing close to the unix epoch. Technically not a problem, but its a bit more aesthetic if they are closer to now, as requested in #10759.

I set them arbitrarily to two hours in the past so that they do not run immediately into the future.

[skunert]

/cmd prdoc --audience node_dev --bump patch

[bkchr]

I find 2 hours in the past also random, but whatever :P

[skunert]

/cmd fmt

[paritytech-workflow-stopper]

All GitHub workflows were cancelled due to failure one of the required jobs.
Failed workflow url: https://github.com/paritytech/polkadot-sdk/actions/runs/21028350688
Failed job name: build-templates-node

[pgherveou]

This PR introduced a DX regression for runtimes that use RelaychainDataProvider as the scheduler's BlockNumberProvider (which includes all system parachains).

The problem: Before this PR, relay_offset defaulted to 0, so relay_parent_number started near 0 and incremented by ~2 per para block. After this PR, relay_offset is derived from wall clock time ((unix_now - 2h) / 6000), which is ~295M in Feb 2026.

pallet_scheduler::service_agendas() iterates from Scheduler::IncompleteSince up to now (the relay block number), checking each relay block's agenda. Since IncompleteSince starts near 0 but now jumps to ~295M, the scheduler burns its entire weight budget (~8.3MB proof_size) doing 197 empty agenda lookups per block, consuming 85.7% of the max PoV budget on every idle block.

Workaround: inject both Scheduler::IncompleteSince and ParachainSystem::LastRelayChainBlockNumber into the raw genesis spec, set to the expected initial_relay_slot value. This closes the gap so the scheduler only does 1-2 iterations per block.

We should work on a proper fix to make polkadot-omni-node easier to use in dev mode without requiring to tweak the genesis.

[skunert]

My initial impression is that this should be fixed at the scheduler level. This comment already foresees this issue:

polkadot-sdk/substrate/frame/scheduler/src/lib.rs

Line 327 in 2de7352

/// *Agile Coretime* chain with relay chain number provider configured, it could otherwise

Main question for me is why Scheduler::IncompleteSince is even initialized with a low value. Dev chain is started and the first thing the runtime sees should already be a high block number. How does such a big gap even get created?