Merge pull request #5137 from shuujii/use-uintptr_t-instead-of-unsigned-long-in-include-mruby-boxing_word.h

Use `uintptr_t` instead of `unsigned long` in `include/mruby/boxing_word.h`

Avoid undefined behavior

Simplify `obj_free`

### ASAN report (`MRB_INT32`)

  ```console
  $ bin/mruby -ve '-0x40000000'
  mruby 3.0.0preview (2020-10-16)
  00001 NODE_SCOPE:
  00001   NODE_BEGIN:
  00001     NODE_NEGATE:
  00001       NODE_INT 40000000 base 16
  irep 0x6070000001e0 nregs=2 nlocals=1 pools=0 syms=0 reps=0 iseq=9
  file: -e
  /mruby/src/codedump.c:173:49: runtime error: left shift of 49152 by 16 places cannot be represented in type 'int'
  SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior /mruby/src/codedump.c:173:49 in
      1 000 OP_LOADI32	R1	-1073741824
      1 006 OP_RETURN	R1
      1 008 OP_STOP

  /mruby/src/vm.c:1138:7: runtime error: left shift of 49152 by 16 places cannot be represented in type 'mrb_int' (aka 'int')
  SUMMARY: UndefinedBehaviorSanitizer: undefined-behavior /mruby/src/vm.c:1138:7 in
  ```

`mrb_bool`  on LLP64 environment seems to become false if, for example,
`mrb_value` is a pointer whose lower 32 bits are 0.

Fix paths of gem files to be scanned for presym

Fix that large integer is not GCed with Word-boxing

When building from outside `MRUBY_ROOT` (e.g. `rake -f`), gem files were not
scanned.

### Example (32-bit Word-boxing)

  ```ruby
  # example.rb
  int_count = ObjectSpace.count_objects[:T_INTEGER]||0
  int = 1<<30
  p (ObjectSpace.count_objects[:T_INTEGER]||0) - int_count
  int = nil
  GC.start
  p (ObjectSpace.count_objects[:T_INTEGER]||0) - int_count
  ```

#### Before this patch:

  ```console
  $ bin/mruby example.rb
  1
  1
  ```

#### After this patch:

  ```console
  $ bin/mruby example.rb
  1
  0
  ```

Fix overflow detection in integer operations with `MRB_WORD_BOXING`.
This bug made `1073741824 == 1073741824+0` to be `false` on 32bit
platforms.

Should raise `RangeError` if the operation overflows.

Remove redundant command execution in `header_search_paths`

There is no need to execute command in duplicate for the same
reason as #5128.

Remove unused constant in `lib/mruby/build/command.rb`

Workaround to build `mirb` with Homebrew's readline on Mac; fix #4537

Simplify `MRuby::Command#_run` to avoid duplicated compilation

Don't use private structs with `mrb_mt_foreach()`

The `MRB_API` function `mrb_mt_foreach()` previously used the private structure `struct mt_elem`.
Therefore, use `mrb_method_t` instead.

ref: https://github.com/mruby/mruby/pull/4959#discussion_r402086196

Compiles twice because it falls back to `build.filename(command)` when
`command` fails. This process was added at 9968af45 to support `ccache gcc`
etc. At that time, it seems that it was necessary because
`build.filename(command)` quoted the whole `command`, but now it does not
quote, so we can just run `build.filename(command)`.

### Example

  ```console
  $ echo 1 > src/a.c
  $ rake -v
  ```

#### Before this patch:

  ```console
  (snip)
  gcc -std=gnu99 -g -O3 -Wall -Wundef -Wdeclaration-after-statement -Werror-implicit-function-declaration -Wwrite-strings -I"/mruby/mruby/include" -MMD -o "/mruby/mruby/build/host/src/a.o" -c "/mruby/mruby/src/a.c"
  /mruby/mruby/src/a.c:1:1: error: expected identifier or '('
  1
  ^
  1 error generated.
  gcc -std=gnu99 -g -O3 -Wall -Wundef -Wdeclaration-after-statement -Werror-implicit-function-declaration -Wwrite-strings -I"/mruby/mruby/include" -MMD -o "/mruby/mruby/build/host/src/a.o" -c "/mruby/mruby/src/a.c"
  /mruby/mruby/src/a.c:1:1: error: expected identifier or '('
  1
  ^
  1 error generated.
  rake aborted!
  (snip)
  ```

#### After this patch:

  ```console
  (snip)
  gcc -std=gnu99 -g -O3 -Wall -Wundef -Wdeclaration-after-statement -Werror-implicit-function-declaration -Wwrite-strings -I"/mruby/mruby/include" -MMD -o "/mruby/mruby/build/host/src/a.o" -c "/mruby/mruby/src/a.c"
  /mruby/mruby/src/a.c:1:1: error: expected identifier or '('
  1
  ^
  1 error generated.
  rake aborted!
  (snip)
  ```

Merge pull request #5127 from shuujii/include-size-of-iv-table-in-ObjectSpace.memsize_of-to-Hash-object

Include size of iv table in `ObjectSpace.memsize_of` to `Hash` object

Use `enable_sanitizer` to simplify `build_config/clang-asan.rb`

Rename `mrb_os_memsize_of_hash_table` to `mrb_hash_memsize`

* The term `hash_table` can be misleading because the return value of this
  function includes memory usage of entire `Hash` object, including not only
  hash table part but also entry list part, etc.
* This function takes a `Hash` object as a receiver and is defined in
  `src/hash.c`, so it is natural to have a `mrb_hash_` prefix.

Change name and usage of presym macros

To be also able to build mruby without presym in the future. However,
`MRB_QSYM` has been removed and changed as follows:

### Example

|       Type                | Symbol |  Previous Style  |   New Style    |
|---------------------------|--------|------------------|----------------|
| Operator                  | &      | MRB_QSYM(and)    | MRB_OPSYM(and) |
| Class Variable            | @@foo  | MRB_QSYM(00_foo) | MRB_CVSYM(foo) |
| Instance Variable         | @foo   | MRB_QSYM(0_foo)  | MRB_IVSYM(foo) |
| Method with Bang          | foo!   | MRB_QSYM(foo_b)  | MRB_SYM_B(foo) |
| Method with Question mark | foo?   | MRB_QSYM(foo_p)  | MRB_SYM_Q(foo) |
| Mmethod with Equal        | foo=   | MRB_QSYM(foo_e)  | MRB_SYM_E(foo) |

This change makes it possible to define, for example, `MRB_IVSYM(foo)` as
`mrb_intern_lit(mrb, "@" "foo")`, which is useful if we support building
without presym in the future.

Remove unused argument of `mrb_obj_iv_tbl_memsize`

Reduce memory usage of Hash object

Fix typo in backward compatible mrb_exc_new_str_lit() macro