Strict validation of AssemblyScript array length

[leoyvens]

See #2140, we know that there are subgraphs in the hosted service that trip this, we should investigate why and fix the root cause. And eventually move to a stricter check.

[evaporei]

Just to clarify things a bit for future use:

• Unfortunately we did not record the information of which subgraph caused this issue;
• The error was a panic when switching chunks for chunks_exact;
• Using chunks might leave unread bytes at the end, which is weird. But we haven't seen any real consequence of this yet.