Inconsistency related to L1_GAS_PER_PUBDATA_BYTE

[mationorato]

First read: matter-labs#1017 (comment)

Better explained by the linked discussion, but tl;dr is that the approach to choosing the cost of L1 gas per pubdata byte appears to be hacky and can cause inconcsistencies.

We should Introduction of a dedicated abstraction as adviced.

[ilitteri]

I think #112 covers this