Add more fine grained options to control the QML disk cache

Currently you can use QML_DISABLE_DISK_CACHE and QML_FORCE_DISK_CACHE to switch any "caching" on and off. This affects the run time caching to .qmlc files as well as the loading of AOT-compiled compilation units.

There are people who really want to use the compiled-in compilation units, but not the caching to .qmlc files. This is important if the file system is constrained in some way.

There should be a way to control this in a more fine grained way:

• Use AOT compiled byte code {yes|no}
• Use AOT compiled C++/machine code {yes|no}
• Use run time disk cache {yes|no|read only}

There is also QV4_FORCE_INTERPRETER that disables the AOT compiled C++/machine code as well as the JIT. And there are the JIT options that handle the decision on whether to use the interpreter or the JIT on any byte code loaded or compiled.