Define subnet type to appropriate size (uint8/uint16)

[y0sher]

Description

Since we only have 126 subnets, there's no need to use large integer to identify subnets, currently uint64 is used which is an overkill. we should consider using something smaller. Even when forward thinking about a future with more subnets, there are smaller integer that will fit.

It will also be easy to define a type Subnet that will be used codebase-wide instead of using some integer, that way it'll be easier if we want to make a change in the future.