Problem:
vim.mpack.encode() picks the msgpack size of a negative integer from its
low 32 bits only, so -(2^32 + 5) is written as a negative fixint and
decodes as -5. Numbers beyond +-2^53 trip an assertion in debug builds.
In release builds, numbers below -2^63 or from 2^64 up go through
out-of-range conversions: -1.5 * 2^63 decodes as 2^62.

Solution:
Use int64 whenever the high word is not a sign extension. Encode numbers
outside [-2^63, 2^64), including NaN and infinities, as floats.

AI-assisted

Problem:
The magic globals `it`, `describe`, etc., are more trouble than they are
worth.

- Hooking into `after_each` requires `getfenv()` hacks.
- They confuse luals/emmylua, because the top-level `.luarc.json` isn't
  merged with `test/.luarc.json` (apparently a luals limitation?)
- They totally defeat discoverability because the user just has to
  "know" about the various magic symbols.

So they harm DX, which means they serve no purpose at all.

Solution:
- Expose the test API from `testutil`, so tests can call `t.it()`,
  `t.describe()`, etc., in the conventional way.
- Drop `getfenv()` hacks.
- Drop the `setfenv()` injection in `load_chunk`.
- Drop `test/_meta.lua`.

Problem:
libmpack encodes boundary values -129 and -32769 with wrong integer
sizes:
- -129 as int8 instead of int16
- -32769 as int16 instead of int32
because the boundary checks compare against the wrong values (e.g., lo
< 0xffffff7f instead of lo < 0xffffff80). This caused data corruption:
-129 would decode as 127.

Solution:
Fix off-by-one errors in the two's complement boundary constants:
0xffffff80 (-128, min int8) and 0xffff8000 (-32768, min int16).

Fixes #37202

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

Problem:
MessagePack fixstr format supports string lengths 0-31, but mpack_str()
only used fixstr for lengths < 20. Strings of 20-31 bytes were
incorrectly encoded as str8 (2-byte header) instead of fixstr (1-byte
header).

Solution:
Change the condition from `len < 20` to `len < 32` to match the
MessagePack specification.

This fix affects message timing which exposed a pre-existing race
condition in the channels_spec PTY test. The test now uses a helper
function to accumulate partial PTY reads, making it more robust.

Fixes #32784

Specifically, functions that are run in the context of the test runner
are put in module `test/testutil.lua` while the functions that are run
in the context of the test session are put in
`test/functional/testnvim.lua`.

Closes https://github.com/neovim/neovim/issues/27004.

ref #24572

Work on https://github.com/neovim/neovim/issues/27004.