Reevaluate EpochID encoding

[ezdac]

Following up on #239 @jannikluhn and I had the discussion whether
the <uint32 block_number><uint32 batch index> coding of the EpochID is necessary.
We think that the EpochID can be any arbitrary number decided upon by the collator, as long as they stay unique (no reuse).
We did not see disadvantages to simply set the EpochID as the L2 BatchIndex, but thought this makes things easier.

Any objections? Why was the block number encoded in the epoch-id?

[ezdac]

#270 could be a reason to keep the combined encoding - but this only makes sense if the user signs the EpochID instead of BatchIndex and L1BlockNumber. Also we have the disadvantage of only 32 bit number-space each and probably would have to migrate to big integers.
I think this is not reason enough, since the combined encoding is not intuitive.

[jannikluhn]

Done in 291.