Fix moving GC and rooting issues with ValueCloner

[Yves]

The ValueCloner in ScriptInterface.cpp implements caching of cloned values by associating a GCThing pointer from another compartment with the cloned version of that value in the current compartment.

This fails with a moving GC because the GCThing pointer can change and because the jsval it uses as value in the map can also change.

There's some more information about moving GC in #2415.

Proposed solution 1

At the moment I don't see why we even need this ValueCloner. I've replaced it with a much simpler implementation with structured clones. ValueCloner was probably introduced earlier when structured clones weren't available yet or when there were some issues with them.

This solution should work with a moving GC and doesn't require special care for rooting (CloneValueFromOtherContext still returns a jsval in this patch, but that should be easy to fix later). CloneValueFromOtherContext is currently mainly used for getting data from the simulation to the GUI. There's another use of it for the serialization test mode in Simulation2.cpp.

Of course when talking about removing something that does caching, the first concern is performance. Since a new instance of ValueCloner is created for each call to CloneValueFromOtherContext, it only caches objects that are linked multiple times in the value to clone. The structured clone most likely does this too internally (otherwise it would get problems with cyclic references).

I've tested it in practice by profiling a game against the AI on Alpine Valleys. Because this cloning is mainly used in the GUI, I couldn't use a non-visual replay. This means the graphs are from different games.

Original:

Patch:

[Yves]

Removes ValueCloner and replaces it with a much simple implementation using StructuredClones

[Yves]

Profiling after applying the patch

[Yves]

Profiling original

[Yves]

In 14724:

Removes ValueCloner.

Fixes #2416
Refs #2415
Refs #1886

[Yves]

I'm reopening this ticket because there are problems with cyclic references which causes the tests to fail at the moment (I don't know if we try to clone any cyclic references somewhere else). Cyclic references work with structured clones in SpiderMonkey 24, but I have to fix the tests because ​sharp variables were ​removed. I will probably revert r14724 and add it to my WIP patch in #1886 together with the fixed tests.

[Yves]

In 14733:

Reverts r14724. Structured clones don't support cyclic references in SpiderMonkey v1.8.5.
SpiderMonkey v24 supports it and I'll add this change to #1886 instead.

Refs #2416
Refs #2415
Refs #1886

[Yves]

In 14877:

Second (main) commit for the SpiderMonkey upgrade.

This commit contains all the required changes to our source files and build scripts (hopefully).
A next commit will remove the old stuff of SpiderMonkey 1.8.5.

Spcial thanks to:

• H4writer who helped a lot mainly with the performance issues we had/have, but also with other problems or questions.
• Leper for the review.
• Historic_bruno for implementing the build scripts on Mac OS X and testing on the Mac.
• The people from the #jsapi channel and from mozilla.dev.tech.js-engine who answered a lot of questions and helped solving problems.
• All the other people who helped

Refs #1886
Fixes #2442
Fixes #2416