[PATCH] Remove unneeded ReserveScriptBackref from the StdDeserializer

[Itms]

If I'm not mistaken, the ReserveScriptBackref system in the deserializer is useless. One must pay attention to creating the backref at the right moment (see r16959) but it's not needed to actually assign the backref after the object is populated.

I think this system was actually useful before we had exact stack rooting.

This patch also contains whitespace fixes, and it makes one of the tests more logical.

[wraitii]

Seems to work fine, can be committed imo.

[Itms]

I'll take care of committing after taking one more look - also I thought I had put leper in CC (since we worked together on fixing bugs here), I'm doing it now.

[leper]

It might have been useful back when it was introcuded (r13429), but the SpiderMonkey API changed since then, and it doesn't seem useful any more. We might create backrefs to something and then fail to actually construct the object, but in those cases we already throw (and we never attempt to recover from that so no real issue there).

You should also nuke FreeScriptBackrefs which is useless since r16214.

Didn't test it though.

[elexis]

Moving to the new WIP milestone.

[Itms]

In 18993:

Remove some obsolete backref logic from the deserializer. It was introduced in earlier versions of the SpiderMonkey API.
Name some variables more logically in one test.

Fixes #4244

[Itms]

Thanks for noticing the FreeScriptBackrefs function.