Lower memory spike when loading with ISQ on CUDA

[EricLBuehler]

Currently when loading with ISQ, there is a spike in GPU memory usage. This is because tensors are copied to the GPU asynchronously and quantized. By forcing a GPU <> CPU synchronization, we can ensure that there is no overlap of operations and that copies are completed, meaning that the spike should be reduced.

[github-actions]

Code Metrics Report

  ===============================================================================
 Language            Files        Lines         Code     Comments       Blanks
===============================================================================
 Dockerfile              1           34           25            0            9
 Happy                   1          442          369            0           73
 JSON                    9           21           21            0            0
 Python                 31         1217         1038           37          142
 TOML                   16          440          400            1           39
-------------------------------------------------------------------------------
 Jupyter Notebooks       1            0            0            0            0
 |- Markdown             1           60           30           22            8
 |- Python               1           96           87            1            8
 (Total)                            156          117           23           16
-------------------------------------------------------------------------------
 Markdown               16         1135            0          836          299
 |- BASH                 5          100           97            0            3
 |- Python               6          122          110            0           12
 |- Rust                 2           80           72            3            5
 (Total)                           1437          279          839          319
-------------------------------------------------------------------------------
 Rust                  115        34379        31132          584         2663
 |- Markdown            57          643           13          596           34
 (Total)                          35022        31145         1180         2697
===============================================================================
 Total                 191        37668        32985         1458         3225
===============================================================================