Blender Git Loki

Manual scaling.

Cleanup: Move curve draw cache implementation to C++

I'd like to use this file to draw curves from geometry nodes, which
would otherwise require implementing a C API. The changes in this
commit are minimal, mostly just casts and changing to nullptr.

Differential Revision: https://developer.blender.org/D11350

Dynamic api switching.

Merge branch 'blender-v2.93-release'

Fix T87357: Missing update after removing socket

The CoW copy of the node group was not updated correctly after it
changed. Arguably, tagging the node tree could also be part of
`ntreeUpdateTree` (which needs to be called after changes to the
node tree anyway). However, there are many instances where the
depsgraph is tagged explicitly after `ntreeUpdateTree` is called,
so it should be fine here as well. This is similar to what is done
in `snode_dag_update`.

Differential Revision: https://developer.blender.org/D11342

Cycles X: Experiment with tile reschedule heuristic

The idea is to add new tiles for rendering when the GPU starts to feel
hungry (as opposite of previous logic which was adding new work tiles
once the number of paths goes below certain threshold). Some motivation
behind this decision:

- There is only that many threads the GPU has. Having much more active
threads might avoid some scheduling latency, but there is limit to how
much it helps.

- Scheduling new tiles early on might have negative effect on coherency,
so allowing more paths to be terminated before re-scheduling keeps the
wavefront more coherent and efficient to be calculated.

The new code will use maximum number of threads the GPU has.

```
new old(1) cycles-x(2) megakernel(3)
bmw27.blend 10.2251 10.198 10.7419 10.4269
classroom.blend 15.8454 16.7821 17.2907 16.6609
pabellon.blend 9.34677 9.39898 9.61772 9.14966
monster.blend 10.374 10.5923 10.5886 12.0106
barbershop_interior.blend 11.5124 11.777 11.8522 12.5769
junkshop.blend 15.6783 16.085 16.2821 16.5213
pvt_flat.blend 16.3432 16.5704 16.2637 17.4047

[1] cycles-x branch, previous commit e0716af1a4f
(2) cyclex-x branch hash ad81074fab1
(3) cyclex-x branch hash ef6ce4fa8ca (right before disabling megakernel)
```

Fix T88375: Bone Size Small After V3D.View_All

The wrong matrix function was used and overwrote the custom bone shape
scale instead of reading from it.

Reviewed By: sybren

Differential Revision: https://developer.blender.org/D11330

Fix T88384: Improved Win32 Window Sizing and Positioning

When creating Win32 windows, the sizes and placements can be out by a
small amount, mostly noticeable near monitor edges. This is because
Windows 10 includes a thin invisible border (typically 7 pixels) when
determining position. Therefore the correct values can sometimes be
just outside the monitor bounds, but we clamp them at those bounds.

This patch fixes this by first clamping the requested values to monitor
bounds, adjusting for window chrome with AdjustWindowRectEx(), and then
using those adjusted values in CreateWindowExW().

see D11314 for more details.

Differential Revision: https://developer.blender.org/D11314

Reviewed by Ray Molenkamp

GPUNodeTree: Move closure socket implicit cast to ntreeGPUMaterialNodes

This reduces complexity of weighting closure in further dev.

GPUNode: Remove ssr_id and sss_id

These are not useful anymore.

Fix particle ID init not creating particle deflect data.

This data (the force fields) are expected to always be there, but they
are currently created on the fly by RNA accessors (typically from UI
draw code), which is extremely wrong way to do it.

Differential Revision: https://developer.blender.org/D11341

Fix some RNA physics accessors creating data.

Accessing data through RNA should never implicitely create it. Objects'
and particles' forcefields and collision settings were doing this.

Note that UI code also had to be tweaked to properly handle `None`
(NULL) cases.

Differential Revision: https://developer.blender.org/D11341

Merge remote-tracking branch 'origin/master' into asset-browser-poselib

GPencil: Speed up Occlude Eraser

This is an initial change to speed up the calculation of the Occlude eraser. In the future, we can add more optimizations, but at least this increase speed.

Intead to check always the 3 points, the check is skipped if it's not required.

Base in a solution by Philipp Oeser.

This is related to T88412

Merge branch 'master' into eevee-gpencil

Merge branch 'master' into greasepencil-object

Cycles: Making sure render is updated when texture options change.

Cycles X: Align kernels of existing and new paths

Only enqueue new kernels when the existing wavefront is at the
intersect closest stage. This seems to positively affect on the
coherency, gaining performance:

```
new cycles-x(1) megakernel(2)
bmw27.blend 10.198 10.6995 10.4269
classroom.blend 16.7821 17.2352 16.6609
pabellon.blend 9.39898 9.65984 9.14966
monster.blend 10.5923 10.5799 12.0106
barbershop_interior.blend 11.777 11.8852 12.5769
junkshop.blend 16.085 16.2971 16.5213
pvt_flat.blend 16.5704 16.3189 17.4047

(1) cyclex-x branch hash ad81074fab1
(2) cyclex-x branch hash ef6ce4fa8ca (right before disabling megakernel)
```

While the pvt_flat (with adaptive sampling) is 1% slower, some
other scenes has performance gained almost all the way back in
comparison to the Cycles-X before megakernel removal.

Note that coherency is a hypothesis. Performance gain might also be
caused by less active paths array calculations.

Cleanup: use nullptr

Cleanup: spelling

Includes fixes to misspelled function names.

Ref D11280