[Bug] Add GC memory pressure tracking for native allocations to avoid memory leaking

[Nucs]

Summary

Fixes #501 and implements #605.

• Add GC.AddMemoryPressure() when allocating native memory via NativeMemory.Alloc
• Add GC.RemoveMemoryPressure() when freeing it
• Update documentation to reflect the new behavior

Problem

When creating many NDArrays in a loop, memory would grow to 10GB+ before GC kicked in:

for (int i = 0; i < 1_000_000; i++)
{
    NDArray array2 = np.array(new double[110]); // 880 bytes each
}
// Before: peaks at 10+ GB
// After: stable at ~54 MB

Root Cause

NumSharp allocates array data via NativeMemory.Alloc (unmanaged) but did not inform the GC about this memory. The GC only saw small managed wrappers (~100 bytes) and was unaware of the ~880+ bytes of unmanaged data per array.

Solution

Track memory pressure in UnmanagedMemoryBlock<T>.Disposer:

Allocation Type	Tracks Pressure?	Reason
Native (NativeMemory.Alloc)	✅ YES	NumSharp allocates → NumSharp tracks
External with dispose	❌ No	Caller allocates → Caller's responsibility
GCHandle (pinned managed)	❌ No	GC already knows about managed arrays
Wrap (no ownership)	❌ No	Not our memory

Test Results

Scenario	Before	After
np.array() × 1M (110 doubles)	~10 GB	54 MB
np.array() × 100K (10K doubles)	1,193 MB peak	46 MB

Checklist

• Native allocations track pressure
• External Disposer has optional bytesCount parameter (default 0)
• Documentation updated