Add More Operations on the Linear Memory

[florianhartung]

Currently, users can only perform byte-wise accesses on the linear memory or borrow the entire linear memory as a mutable slice.

We should add new methods to load/store entire blocks of bytes (e.g. integers).

Should these accesses potentially tear?
Do we still need to acquire a write lock of the linear memory?

The API is not very clear yet and may change with the implementation of shared linear memories.