Rework initial shard slot (PHASE_1_FORK_SLOT)

[hwwhww]

Issue

Fix #1877

How did I fix it

1. Rename PHASE_1_GENESIS_SLOT to PHASE_1_FORK_SLOT: IMHO it's clearer. (Sorry!)
2. Set initial shard_state.slot to compute_previous_slot(pre.slot), where pre is the beacon state.
3. Per @mkalinin's suggestion, make PHASE_1_FORK_SLOT = GENESIS_SLOT = 0 for the test vectors generation.
4. Related to Enable test generation for phase1 #1957 - now we can enable more phase 0 tests to be phase-1-compatible!

TODO

Since it makes PHASE_1_FORK_SLOT = GENESIS_SLOT = 0, we have to add fork boundary tests with a new config with a reasonable value, e.g., PHASE_1_FORK_SLOT := Slot(256). Leave it to future PRs.

[mkalinin]

I have made a couple of suggestions that IMHO are to increase readability for a little bit.

The other thing might worth mentioning. These changes doesn't allow for shard blocks production during PHASE_1_FORK_SLOT while in the case where PHASE_1_FORK_SLOT > GENESIS_SLOT, e.g. the Mainnet, there would be an opportunity to do that as compute_previous_slot(pre.slot) in shard states initializer would return pre.slot - 1.

Do we want to use this opportunity? If yes then PHASE_1_FORK_SLOT in the places where it addresses off-by-one issue should be replaced with GENESIS_SLOT.

[hwwhww]

Note that with this PR:

1. The PHASE_1_FORK_SLOT == Slot(128) case is like:

2. the PHASE_1_FORK_SLOT == GENESIS_SLOT case is like:

The shard block 0 doesn't exist in this case so transition_to_valid_shard_slot helper is for forwarding to Slot(1).

[djrtwo]

excellent!