Replace linear search for free sample index with free list. This brings sound sample index allocation down to constant complexity independent of the number of allocated sounds. On average the time to allocate sound samples is reduced by around 75% (843µs down to 223µs). For perspective, the time to load all default sounds on client launch is reduced by around 15ms (although this does not significantly affect launch time due to threaded loading).
The lock needs to be owned when accessing the sound voices. Calling `IsPlaying` is redundant, as the loops effectively check whether the sample is playing.
Ensure that the UI mouse position stays inside the UI screen rect. Previously, the mouse was not considered to be inside the UI screen when all the way at the right or bottom edge. This caused the map editor tooltip to not be shown when the mouse is all the way at the right side of the map editor view. This is a cleaner fix for #4553 which reverts the previous workaround from #6423.
Add dropdown menus for changing animation and grid settings (i.e. animation speed and grid size) instead of conditionally showing more buttons for this directly in the menu bar. This frees up space in the menu bar, which is currently full on 5:4 resolutions.
The icons previously used for the default animation/grid buttons are now used for the main buttons that toggle animation/grid instead of using text.
Support setting lower animation speeds with the plus and minus buttons by adjusting the step size when the animation speed is low.
Support setting arbitrary animation speed by text input.
Render one quad for each line of the text selection instead of rendering one quad per selected character.
This increases the average FPS when the console is open and all text is selected by around 10% (from around 849 to around 943 FPS) (on my machine, in release mode).
When normalizing color components in the engine graphics, round the components to the nearest integer instead of rounding down. Otherwise the color that is rendered in color pickers may be off by 1 in any of its RGB components from the color that the color picker displays as text (hex string and individual components). The slightly incorrect color can be confirmed by creating a screenshot or otherwise reading the backbuffer (planned editor pipette feature).
This should not change map rendering, since maps already store quantized RGBA values on which the rounding mode should have no effect. It may however slightly change appearance of colors in all other places (at most +1 in every RGB component).
The individual color components were rounded down when being displayed on their own in color pickers, whereas the components are rounded to the nearest integer when packing the colors into the hex string. This was causing minor discrepancies between the color being displayed/saved as hex and the individual components. Rounding the components when packing is necessary to reduce the error when converting between color spaces.
The text color was previously set and reset for every individual number being rendered for non-empty tele, speedup, switch and tune tiles. The color is the same for all tiles from each entities layer, so most of these `TextColor` calls are unnecessary. Now the text color is only set and reset once when rendering each entities layer.
Let's say you have this bind:
```bind x +toggle cl_dummy_hammer 1 0```
and you set cl_dummy_control to 1.
When you press the bind i mentioned above, and then release, the dummy will hammer where he is looking (not at you). So, in total, there will be two hammers. One hammer when you press down the button and the dummy hammers towards you, and then another hammer when you release the button and the dummy hammers where he is looking.
This fixes it, and also makes sure it does not conflict with cl_dummy_copy_moves (as if it is enabled and cl_dummy_control is enabled, the dummy will not copy fire, hook, or jump) so I made sure it keeps this functionality as it's pretty cool.
This does not fix any other bugs yet, maybe I will fix those in the future but we'll see. Any bug you may encounter with this change is also probably present in the main branch, such as resetonswitch not working perfectly with dummy_control, but if you do find something different then let me know.