Add async kernel execution API

[gshaowei6]

Fixes #41.

This adds a small asynchronous execution handle for kernels:

• Kernel.executeAsync(...) overloads for Range, int, passes, and explicit entrypoints
• public Execution handle with waitUntilFinished(), isFinished(), and getKernel()
• background execution delegates to the existing synchronized execute(...) path, preserving current execution behavior while letting callers wait explicitly

Validation:

• RED: KernelAsyncExecutionTest initially failed to compile because com.aparapi.Execution and Kernel.executeAsync(Range) did not exist.
• Focused compile with javac -g --release 8 for Execution, Kernel, and KernelAsyncExecutionTest.
• JUnitCore com.aparapi.runtime.KernelAsyncExecutionTest com.aparapi.runtime.MultiplePassesMemoryConsumptionTest passed (2 tests).
• git diff --check passed.