Implemented a Flexible and Adaptive Staging System

[daniel86]

This is a larger PR where I migrated code to immutable storage API, and introduced a staging system for centralized memory management for writing and reading data from GPU memory.

• Summarizing the staging system manages multiple semantic buffer ranges within a buffer dedicated to to CPU-based access of GPU memory. The BOs are organized by size and update frequency and adaptively moved to the most apropiate staging arena. Some of the arena types use ring buffers whose size also can adaptively change depending on stall rate when waiting for CPU-GPU synchronization.
• Staging system supports UBO, SSBO and TBO
• Further introduced a thread-safe ClientBuffer class that can be used to read/write data which is added into the staging system. The client buffer also supports an adaptive mode where it switched to double buffering in case of wait delays. This can be used to insert data safely from different threads into staging system and upload to GPU
• Notably I also introduced support for indirect multi-layer drawing

[github-actions]

✅ Compilation - SUCCESS! ✅

[github-actions]

Test Results

1 tests  ±0   1 ✅ ±0   0s ⏱️ ±0s
1 suites ±0   0 💤 ±0 
1 files   ±0   0 ❌ ±0 

Results for commit c1d0768. ± Comparison against base commit cfd83d8.