Fixed CUBLAS DLL load issues on Windows

[aneeshjoy]

Description:

This change fixes DLL load issues when CUBLAS is enabled on Windows.

Symptoms:

Traceback (most recent call last):
  File "path redacted...\venv\lib\site-packages\llama_cpp\llama_cpp.py", line 55, in _load_shared_library
    return ctypes.CDLL(str(_lib_path))
  File "C:\Program Files\WindowsApps\PythonSoftwareFoundation.Python.3.10_3.10.3056.0_x64__qbz5n2kfra8p0\lib\ctypes\__init__.py", line 374, in __init__
    self._handle = _dlopen(self._name, mode)
FileNotFoundError: Could not find module 'path redacted...\venv\Lib\site-packages\llama_cpp\llama.dll' (or one of its dependencies). Try using the full path with constructor syntax.

Root cause:
Python for Windows does not search for DLLs in directories listed in PATH environment variables. So CUDA libraries are not available while loading llama.dll. Hence the error module or dependencies not found.

Python change log for Windows

Fix:
Add CUDA_PATH to dll search directory via add_dll_directory

Result:

llama.cpp: loading model from .....redacted.....
llama_model_load_internal: format     = ggjt v2 (latest)
llama_model_load_internal: n_vocab    = 32001
llama_model_load_internal: n_ctx      = 512
llama_model_load_internal: n_embd     = 4096
llama_model_load_internal: n_mult     = 256
llama_model_load_internal: n_head     = 32
llama_model_load_internal: n_layer    = 32
llama_model_load_internal: n_rot      = 128
llama_model_load_internal: ftype      = 2 (mostly Q4_0)
llama_model_load_internal: n_ff       = 11008
llama_model_load_internal: n_parts    = 1
llama_model_load_internal: model size = 7B
llama_model_load_internal: ggml ctx size =  72.75 KB
llama_model_load_internal: mem required  = 5809.34 MB (+ 1026.00 MB per state)
llama_model_load_internal: [cublas] offloading 24 layers to GPU
llama_model_load_internal: [cublas] total VRAM used: 2895 MB
llama_init_from_file: kv self size  =  256.00 MB
AVX = 1 | AVX2 = 1 | AVX512 = 0 | AVX512_VBMI = 0 | AVX512_VNNI = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 1 | SSE3 = 1 | VSX = 0 | 

[abetlen]

@aneeshjoy I see the python changelog, that makes sense, however is bin the right path? In any case I've merged this, hope it helps.