Timlord updates state immediately, prevents race condition

[mariano54]

• The first issue is what we saw in the first epoch in testnet6. The full node was hardcoding the default difficulty if block height is < EPOCH_BLOCKS. However there were many overlapping blocks, so none of the blocks reached the height, and therefore the timelord infused the wrong difficulty.
• The second fix is fixing a race condition in the timelord, where it took time to update the state, so it ignored the new_peak_timelord form the full_node, which should have reset the timelord to a good state.

This should be tested for a few hours on the fastest timelord in testnet6.

[hoffmang9]

+1