Emit Better Error Message for Function Declarations in Invalid Locations

[IgnoredAmbience]

On the conrad-v8-merged branch (I think), latest revision.
The following test cases aborted with a hard crash from somewhere in the parser or its interface.

Quick diagnosis would be appreciated so I can classify the bug or choose to ignore it for the CAV'15 paper writeup.

	Record 1
id	1753444
test_id	tests/test262/ch10/10.4/10.4.2/10.4.2-1-2.js
batch_id	438447
result	ABORT
exit_code	2
stdout
stderr	Fatal error: exception Parser.InvalidArgument
duration	0

	Record 2
id	1754068
test_id	tests/test262/ch12/12.14/12.14-13.js
batch_id	438603
result	ABORT
exit_code	2
stdout
stderr	Fatal error: exception Parser.InvalidArgument
duration	0

	Record 3
id	1753805
test_id	tests/test262/ch12/12.8/S12.8_A3.js
batch_id	438537
result	ABORT
exit_code	2
stdout
stderr	Fatal error: exception Parser.InvalidArgument
duration	0

	Record 4
id	1753802
test_id	tests/test262/ch12/12.8/S12.8_A4_T1.js
batch_id	438536
result	ABORT
exit_code	2
stdout
stderr	Fatal error: exception Parser.InvalidArgument
duration	0

	Record 5
id	1753807
test_id	tests/test262/ch12/12.8/S12.8_A4_T2.js
batch_id	438537
result	ABORT
exit_code	2
stdout
stderr	Fatal error: exception Parser.InvalidArgument
duration	0

	Record 6
id	1753806
test_id	tests/test262/ch12/12.8/S12.8_A4_T3.js
batch_id	438537
result	ABORT
exit_code	2
stdout
stderr	Fatal error: exception Parser.InvalidArgument
duration	0

[IgnoredAmbience]

There's no other stdout/stderr output collected by the runtests.py script, it could have got lost somewhere, if other output is present on the console, we should log a minor bug against that, too.

[IgnoredAmbience]

This is our old friend of permitted locations for function declaration syntax. (Thanks @Pasteuriser for confirming.)

Classifying as buggy tests.

We should emit a better error message in this instance.

[IgnoredAmbience]

Also tests/test262/ch15/15.4/15.4.4/15.4.4.19/15.4.4.19-5-1.js

[conrad-watt]

One thing I didn't appreciate before; in ES6, this syntax will be standard-compliant, since FunctionDeclarations can now occur in a block (and the declaration is block-scoped). However, there are some tests here from 2009 and possibly earlier, so I don't think this is a case of eager conformance.

[IgnoredAmbience]

Yep, I had in the back of my head that they were going to get rid of these sort of discrepancies between the standard and common use (proto, is another example, I believe)

It was a bug in the test262 es5 suite, and that's how I've classified it...