Fix type instability with high thread counts

[ChrisRackauckas]

Summary

This PR fixes a critical type instability bug that was introduced when replacing CPUSummary.sys_threads() with Threads.nthreads() in PR JuliaSIMD#28.

The Problem

With the current implementation, when using more than 64 threads:

• worker_bits() returns a regular Int64 instead of StaticInt
• This propagates to worker_mask_count() which also returns Int64
• The _request_threads function expects StaticInt{N} but receives Int64, causing a MethodError

The Solution

This PR makes the following changes:

1. Modified worker_bits() to always return Int for consistency
2. Updated worker_mask_count() to use regular integer division (cld)
3. Added a new _request_threads method that handles Int parameters
4. Added tests for high thread count compatibility

Testing

The fix has been tested with:

• 1 thread (works as before)
• 128 threads (now works correctly)

Before the fix:

julia> using Polyester
julia> @batch for i in 1:10; println(i); end
ERROR: MethodError: no method matching _request_threads(::UInt32, ::Ptr{UInt64}, ::Int64, ::Nothing)

After the fix:

julia> using Polyester
julia> @batch for i in 1:10; println(i); end
# Works correctly

Compatibility

The fix maintains backward compatibility while ensuring the code works correctly with any number of threads.

🤖 Generated with Claude Code