Zero-padding in storage should not be applied

[gavofyork]

Currently when reading from storage, if the entry in storage is too small, zeroes are appended. This works fine for single LE ints, but less well in pretty much any other situation and could easily lead to security issues.

Zero-padding should only be applied when reading specific types, essentially just Balance. In other instances, the runtime should simply be allowed to fail.

In any case, all zero-padding this should be removed entirely once we decommission the PoC-1/2 testnet.

[rphmeier]

I don't understand why zero-padding is useful for reading integers either.

[guanqun]

curious to know why we need the zero-padding in the first place. for backward-compatibility?

[gavofyork]

indeed.

[gavofyork]

Closed with #709