Like line2byte, but works for any buffer, and uses zero-based
indexing (API conventions).

msgpack_rpc_to_object (called by handle_request .. msgpack_rpc_to_array)
always NUL-terminates API Strings.
But handle_request .. msgpack_rpc_get_handler_for operates on a raw
msgpack_object, before preparation.

Also re-word some error messages:
- "Key does not exist: %s"
- "Invalid channel: %<PRIu64>"
- "Request array size must be 4 (request) or 3 (notification)"
- "String cannot contain newlines"

References #6150

Should make me able to determine whether they are lua bindings that contain 
a bug or set_lines.

Hope this will make people using feed_command less likely: this hides bugs.
Already found at least two:

1. msgpackparse() will show internal error: hash_add() in case of duplicate
   keys, though it will still work correctly. Currently silenced.
2. ttimeoutlen was spelled incorrectly, resulting in option not being set when
   expected. Test was still functioning somehow though. Currently fixed.

When the buffer that nvim_buf_set_lines() is changing is not in any vim
window, fix_cursor() leads to calling ml_get_buf() with an invalid line
number. The condition that fix_cursor() was called on was (buf ==
curbuf), but this is always true because of the call to
switch_to_win_for_buf() earlier in the function.

Instead this should be predicated on (save_curbuf.br_buf == NULL)