[BOUNTY $20] Support for syncing threads

[CC007]

It would be nice to have the ability to sync kernel threads when the kernel is running (similar to how CUDA has __syncThreads()

[freemo]

@CC007 I have included this in the list of tickets I'll be adding a bounty to as well. Let me know if you have any interest in taking it up.

[freemo]

@CC007 do you think our recent implementation of barriers satisfies this ticket?

[KingParmenides]

Opened PR #193 for this bounty: #193

It adds a CUDA-style syncThreads() helper that maps to barrier(CLK_LOCAL_MEM_FENCE | CLK_GLOBAL_MEM_FENCE) and uses the existing Java barrier path, with codegen and JTP runtime coverage.

Bounty payout addresses if needed:

• BTC: 39Q34P8A7g375yqEr8buvNJkUbRgKfbKQZ
• USDC: 0x9343305D20d12c0aDF97F943bb2152F8eC4Cef4C