We should remove the shims for the LazyScript/JSScript constructor and use a single signature defined on BaseScript by including using BaseScript::BaseScript in sub-classes.

Also should move the lineno/column initialization to js::BaseScript constructor with the rest of the source location information.

The line number information is already available on either CompileOptions or
FunctionBox when we construct the JSScript. Initialize the the fields in the
constructor instead of later on in JSScript::fullyInitFromEmitter.

Directly wrap the js::BaseScript constructor and fix up argument names and
ordering as needed. This moves some behavior from LazyScript constructor to
LazyScript::CreateRaw. The immutableFlags for LazyScript are also initialized
in callers for more consistency with JSScript.

Depends on D56194

Backed out 2 changesets (bug 1601963) for causing web-platform-test assertion failures on JSScript.cpp CLOSED TRE

Backout revision https://hg.mozilla.org/integration/autoland/rev/ec92abf447a66e402d3630ab219cf1f7dfcf29e0

Failed push https://treeherder.mozilla.org/#/jobs?repo=autoland&resultStatus=testfailed%2Cbusted%2Cexception&revision=f8154935d6a2997c178300509eb1185b190d4cc7&searchStr=wpt&selectedJob=280077220

Failure log https://treeherder.mozilla.org/logviewer.html#?job_id=280077220&repo=autoland

Ted can you please take a look?

The JSScript constructor need to use the same initialization as the BCE for the lineno in the BinAST case. I've updated the patch.

https://searchfox.org/mozilla-central/rev/0815480479c19646f9330377ffd8730515fca4ec/js/src/frontend/BytecodeCompiler.cpp#761,787

https://hg.mozilla.org/mozilla-central/rev/40f5ce44be60
https://hg.mozilla.org/mozilla-central/rev/7fd9c7c41a1b