Allow users to specify build optimization level

[nhuang-tt]

Ticket

#7449

Problem description

Compute kernels are hardcoded to O3. Other kernels are hardcoded toOs.
Users can't benefit from potentially "free" optimization wins because the optimization flag to build their kernels is hardcoded to Os.

What's changed

Add opt_level to kernel config structs. Default to O3 (compute) and Os (data movement, ethernet) if not specified.

Example: The user can specify the data movement kernel to be built with O3 optimization. If this flag isn't specified, it will use the default which is Os.

    KernelHandle void_dataflow_kernel_noc0_id = CreateKernel(
        program,
        "tt_metal/programming_examples/hello_world_datamovement_kernel/kernels/dataflow/void_dataflow_kernel.cpp",
        core,
        DataMovementConfig{
            .processor = DataMovementProcessor::RISCV_0,
            .noc = NOC::RISCV_0_default,
            .opt_level = KernelBuildOptLevel::O3
    });

Checklist

• All post commit CI passes --- same as main
  https://github.com/tenstorrent/tt-metal/actions/runs/13518124178
• Blackhole Post commit CI passes (if applicable)
• Model regression CI passes (if applicable)
• Device performance regression CI passes (if applicable)
  https://github.com/tenstorrent/tt-metal/actions/runs/13518134596
• (For models and ops writers) Full new models tests CI passes (if applicable)
• New/Existing tests provide coverage for changes
  TGG: https://github.com/tenstorrent/tt-metal/actions/runs/13518130484

[pgkeller]

one nit