feat: async/parallel execution with configurable concurrency

Parallelize LLM calls across minibatches to reduce wall-clock time.
All domain ports (LLMPort, JudgePort, ProposerPort) are now async.
Adapter implementations wrap synchronous DSPy calls with asyncio.to_thread.
Judge calls run in parallel within a batch using asyncio.gather + semaphore.
Evaluator parallelizes minibatch execution with configurable concurrency.
Evolution loop and use case are fully async. Proposer stays sequential.
Added --max-concurrency CLI flag and max_concurrency YAML config field.
Added async_retry_with_backoff for async error handling.
All 139 unit tests pass.

Co-Authored-By: Paperclip <noreply@paperclip.ing>

src/prometheus/cli/app.py

@@ -5,6 +5,7 @@ Typer interface with -i (input) and -o (output) options.
 """
 from __future__ import annotations
 
+import asyncio
 import logging
 import os
 from dataclasses import asdict
@@ -66,12 +67,30 @@ def optimize(
         "--error-strategy",
         help="How to handle errors: skip | retry | abort.",
     ),
+    max_concurrency: int = typer.Option(
+        5,
+        "--max-concurrency",
+        help="Max parallel LLM calls for minibatch execution and judging.",
+    ),
 ) -> None:
     """Optimize a prompt without any reference data.
 
     Usage:
         prometheus optimize -i config.yaml -o result.yaml
     """
+    asyncio.run(
+        _async_optimize(input, output, verbose, max_retries, error_strategy, max_concurrency)
+    )
+
+
+async def _async_optimize(
+    input: str,
+    output: str,
+    verbose: bool,
+    max_retries: int,
+    error_strategy: str,
+    max_concurrency: int,
+) -> None:
     # Configure verbose logging
     if verbose:
         logging.basicConfig(level=logging.INFO, format="[PROMETHEUS] %(message)s")
@@ -129,6 +148,7 @@ def optimize(
         retry_delay_base=raw_config.get("retry_delay_base", 1.0),
         circuit_breaker_threshold=raw_config.get("circuit_breaker_threshold", 5),
         error_strategy=raw_config.get("error_strategy", error_strategy),
+        max_concurrency=raw_config.get("max_concurrency", max_concurrency),
         output_path=output,
         verbose=verbose,
     )
@@ -164,6 +184,7 @@ def optimize(
         lm=judge_lm,
         max_retries=config.max_retries,
         retry_delay_base=config.retry_delay_base,
+        max_concurrency=config.max_concurrency,
     )
     proposer_adapter = DSPyProposerAdapter(
         lm=proposer_lm,
@@ -171,7 +192,11 @@ def optimize(
         retry_delay_base=config.retry_delay_base,
     )
     bootstrap = SyntheticBootstrap(generator=synth_adapter, seed=config.seed)
-    evaluator = PromptEvaluator(executor=llm_adapter, judge=judge_adapter)
+    evaluator = PromptEvaluator(
+        executor=llm_adapter,
+        judge=judge_adapter,
+        max_concurrency=config.max_concurrency,
+    )
     use_case = OptimizePromptUseCase(
         evaluator=evaluator,
         proposer=proposer_adapter,
@@ -180,7 +205,7 @@ def optimize(
 
     # 4. Execute
     with console.status("[bold green]Evolving prompt..."):
-        result = use_case.execute(config)
+        result = await use_case.execute(config)
 
     # 5. Display results
     _display_result(result)