This PR simplifies and replaces `historical_roots` with
`historical_block_roots`.

By keeping an accumulator of historical block roots in the state, it
becomes possible to validate the entire block history that led up to a
particular state without executing the transitions.

This is interesting for archival purposes as well as when implementing
sync protocols that can then proceed to verify chunks of blocks quickly
- it's also useful as it provides a canonical hash by which such chunks
of blocks can be named, with a direct reference in the state.

In order not to grow the state size further, the `historical_roots`
field is removed - the blocks already contain state roots, so the value
of repeating the state root in the historical accumulator is limited to
empty slots, and even then, the historical block root accumulator
differentiates states that differ by empty slots.

This is a sketch of the full PR - there are a few key decisions to make:

* should the historical block roots be backfilled with phase0 data on
the transition? This simplifies future logic but could be omitted for
simplicity. It's easy to precalculate what the values should be, so it's
not a computational issue.
* should historical roots remain?

As far as naming goes, it's convenient to talk about an "era" being 8192
epoch = 256k slots ~= 36.4 days.

* enables proving the state for empty slots
* allows verifying the backfilled fields against the previous
`historical_roots` field
* allows trivially computing `historical_roots`, for any existing use
cases