Change assert to allow blank strings

[DuncanUszkay1]

Issue: #1176

When using the documented --use abort=abort would be set to an empty string, which resulted in the call to global aliases returning ''. Then since blank strings are falsey, this assertion would fail.

I've changed the assertion to assert that the name is a string, which catches the case that the alias is missing but does not fail in the case of an empty string

examples:
If the name isn't found in the aliases table:

//let alias = unchecked(_keys[i]);
let alias = "not the right key"
let name = globalAliases.get(alias);
assert(isString(name));

We get: ERROR: AssertionError: assertion failed as expected.

In the case where you specify an abort that does not match any export:
ERROR: no such global element: main/nope as before

In the case where you use a blank abort, compilation is successful.

[dcodeIO]

Can just use changetype<string>(globalAliases.get(alias)) since the value can't be null

[DuncanUszkay1]

but it can be undefined right?

alias = "bad key"
let name = changetype<string>(globalAliases.get(alias))
if (!name.length) continue; // explicitly disabled

throws

ERROR: TypeError: Cannot read property 'length' of undefined

I suppose that's not really much different than the assertion though and shouldn't really happen anyway

[dcodeIO]

Hmm, yeah, that's strange. Might be cli/asc.js wrongly inserting an undefined into the map.

[dcodeIO]

Btw, alias = "bad key" doesn't happen since we _keys = Map_keys(globalAliases) so we know the key is in there

[DuncanUszkay1]

Ah okay- well in that case maybe we should just use ! and ignore the undefined case

[DuncanUszkay1]

Not sure I understand why CI is failing here- doesn't appear to be related to the PR

[dcodeIO]

Seems to me that there's something wrong with GitHub's git, hmm. Will try to rerun the CI job at a later time.

[DuncanUszkay1]

Is this blocked on CI failures?

[DuncanUszkay1]

CI is passing now, should be good to merge?