win-capture: Fix leaking framebuffers data

[sumydi]

Description

The wrong linked list was used when removing framebuffer data. It generated memory leak, the linked list grew more and more involving performance degradation each time an application called the vkCmdBeginRenderPass.
Furthermore, few pointers referenced objects on the stack that were no more valid.

Motivation and Context

It solves a memory leak and performance problems.

How Has This Been Tested?

Tested on Windows 10.

Types of changes

• Bug fix (non-breaking change which fixes an issue)

Checklist:

• My code has been run through clang-format.
• I have read the contributing document.
• My code is not on the master branch.
• The code has been tested.
• All commit messages are properly formatted and commits squashed where appropriate.
• I have included updates to all appropriate documentation.

[derrod]

Please adjust the commit message according to our contributing guidelines.

We also have been a bit inconsistent with the prefix for the hook, but it should either be win-capture: (the plugin) or graphics-hook: (for the sub project).

[gxalpha]

You probably want to target the master branch instead of the release/29.1 branch, which is unlikely to get any more releases anyways.
I’ll change the base branch of the PR to master, please make sure to rebase your changes on master (or reset it to the state of master and cherry-pick your commit) and force-push that to the branch of this PR (otherwise the cherry-picked commits on 29.1 will also show up on the PR).

[sumydi]

Yes, you're right, my bad, I was not aware that the 29.1 branch was "dead".
I rebased my changes to master and force-pushed the change to the PR's branch.
I hope that the commit message is now compliant with OBS guidelines.

[derrod]

Commit message is fine now, I'll leave the functional review to somebody whose more familiar with Vulkan stuff.

[jpark37]

Fixes look good. @Lain-B might want to bump the hook version though.

[Lain-B]

Fixes look good. @Lain-B might want to bump the hook version though.

Will do.

[Lain-B]

Thank you very much for the fix.

[sumydi]

You're welcome, I'm happy to have helped.

I don't know if the following information can be useful for you, but:

• the bug can degrades the performances of any application that is using Vulkan each time it calls vkCmdBeginRenderPass()
• the severity of the issue depends on how the application manages the vkFrameBuffer
• one potential workaround for the application is to use the VK_KHR_dynamic_rendering extension instead of vkRenderPass,