Clearing all method cache using `memset` is faster than conditional
clear by method id.

You don't need pointer tweaking on 32bit architecture, where pointers
fit in 32bit (lower half of mrb_value).

Tested only on Linux. You need to install 32bit relate libraries,
e.g. libc6-dev-i386.

On 64bit platforms: `MRB_NAN_BOXING`
On 32bit platforms: `MRB_WORD_BOXING`
On debugging: `MRB_NO_BOXING`

Since it's an internal static function.

When the array is very big, the simpler `mrb_write_barrier` causes
calling `gc_mark_children` for big arrays repeatedly. That would hinder
performance very badly.

Which is caused by `MRB_NAN_BOXING`.

Introduced `MRB_NO_METHOD_CACHE` which is inverse of `MRB_METHOD_CACHE`
that should be enabled intestinally. In addition, the default cache is
made bigger (128 -> 256).

The minor versions should be upper compatible. So mere opcode, section
addition can be done without breaking compiled binary.

- no OP_EXT_ anymore
- OP_LOADI16 in right position

[ruby-bugs:15921]

They are included from `mruby.h` anyway, and including it ahead can
cause some errors regarding `INT32_MAX` etc. with C++ compiler.

This combination is too hard to support. Difference between C and C++ is
too big with VisualC++. GCC and clang are OK.