fix(esx_lib/streaming): give asset requests a timeout

[seltonmt012]

Every request function loops while not HasXLoaded(...) do Wait(500) end and never gives up. A misspelled or unavailable anim dict, texture dict or ptfx asset makes that loop run forever, so the calling coroutine never returns. esx_progressbar waits on RequestAnimDict, so a bad dict there leaves the player stuck. A not yet loaded asset also costs at least 500 ms, since the first check always fails.

dev fixed this in 0f5f6a6 with a waitForLoaded helper: 5000 ms timeout, 50 ms step, and a shortcut when the asset is already loaded. It was not carried over when the module moved into esx_lib, so v1.14.1 and main still run the old code. This ports it and does the same for requestAudioBank, which dev does not have.

Callers already deal with nil. requestModel could always return nil through the IsModelInCdimage guard, and esx_multicharacter and spawnVehicle both check for it.

Measured in game on artifact 25770 with a dict name that does not exist:

before:  never returned, still hanging after 8 seconds
after:   returns nil after 5026 ms

A valid dict still returns immediately.

• Conventional Commits.
• Tested locally.
• No breaking changes.
• Clear explanation.

[N0tNvll]

I suggest implementing a default streaming timeout and allowing an optional timeout for each streaming function

[seltonmt012]

Added. The default stays at 5000 ms and every request function takes an optional timeout after the callback, so existing calls are unchanged.

Measured in game on artifact 25770:

bad anim dict, no timeout arg        nil after 5004 ms
bad anim dict, 1000 ms               nil after 1040 ms
bad anim dict, 200 ms                nil after  231 ms
bad texture dict, 800 ms             nil after  830 ms
valid anim dict, no timeout arg      returned after   0 ms
valid anim dict, 1000 ms             returned after  56 ms
valid anim dict, callback, 3000 ms   callback result after 0 ms
valid model, 4000 ms                 returned after 119 ms