OLMKit

Add some tests for the Javascript wrappers

These would have helped avoid the recent FRV.

Avoid buffer overrun on encryption

Make sure we null-terminate encrypted strings before passing them to
UTF8ToString.

This used to work when we allocated the buffer on the stack, because it turns
out that allocate() zeroinits the returned memory. malloc(), of course, does
not.

Allocate memory for message blobs on the heap

Avoid leaving copies of the plaintext sitting around in the emscripten heap.

Better handling of exceptions during loading, and import OLM_OPTIONS

Messages can be very large, so we don't really want to allocate them on the
stack. Switch to using the heap for them, and try to clean up some of the
string handling while we're at it.

Read settings from OLM_OPTIONS to allow apps to configure some options. In
particular, this is useful for setting the heap size.

If we get an exception during load, don't define half of window.Olm (which
confuses apps).

This is a partial fix to https://github.com/vector-im/riot-web/issues/2726.

OLMKit: Make returned NSError provide the raw olm error string (ex:"UNKNOWN_MESSAGE_INDEX") in their NSLocalizedDescriptionKey.

NSLocalizedFailureReasonErrorKey can contain more contextual information.

OLMKit: Fixed type-cast of messageIndex of [OLMInboundGroupSession decryptMessage] for 32 and 64bits platforms

fuzz_group_decrypt.cpp got broken by 653790ea; fix it up