Wireshark: style tweaks akin to Wireshark practices

[Boolean263]

Some minor tweaks to the coding style of the Wireshark dissector to bring it closer to the standard practices used in the Wireshark project itself. All changes are compatible with Wireshark 4.6 and should be with 4.4 as well.

Define WS_LOG_DOMAIN. Allows for more targeted showing/hiding of log messages. Use ws_warning() instead of g_warning() to leverage.

Correct spelling of MAX_SSL_VERSION() macro.

Run script through Wireshark's tools/convert-glib-types.py to replace GLib-specific types with C99 standard types (ref Wireshark #19116).

Make every global variable and every function (except the register and handoff functions) static, since none of them are used outside of this plugin.

Remove constant HFIDS and instead define interesting_hfids[] as having an indefinite number of elements. Use the array_length() macro instead for determining the size of the array. It's still calculated at compile time, and removes a potential source of bugs if elements are added to or removed from interesting_hfids[] in future changes.

Add g_strfreev() calls to correspond with each call to g_strsplit(), which specifies in its
documentation that its return value must be freed.

[vlvkobal]

Hi David, the tests fail. Please look at

tshark: Dissector bug: register_subtree_array: subtree item type (ett_...) not -1 !
This is a development error: Either the subtree item type has already been assigned or was not initialized to -1.

[Boolean263]

Ah, I see. The tests use Wireshark 4.2.2 which predates the change around initializing static variables. I missed this because I've been doing my in-tree dev work against Wireshark's master branch, which is closest to the Wireshark 4.6.0 release.

That part should be easy enough to revert while keeping the rest of the changes, but I'll test it against 4.2.2 to see if there are any other modernisms that would break that version. (I'm assuming you wouldn't test that version if you were able to migrate to something newer.)