feat(ralph-loop): add strategy option for fresh context per iteration

Closes #1901

Add 'default_strategy' config option (default: 'continue') to control whether ralph-loop creates a new session per iteration ('reset') or keeps the same session ('continue'). The 'reset' strategy keeps the model in the smart zone by starting with fresh context for each iteration.

Supports --strategy flag for per-command override.

src/hooks/ralph-loop/command-arguments.ts

@@ -0,0 +1,27 @@
+export type RalphLoopStrategy = "reset" | "continue"
+
+export type ParsedRalphLoopArguments = {
+  prompt: string
+  maxIterations?: number
+  completionPromise?: string
+  strategy?: RalphLoopStrategy
+}
+
+const DEFAULT_PROMPT = "Complete the task as instructed"
+
+export function parseRalphLoopArguments(rawArguments: string): ParsedRalphLoopArguments {
+  const taskMatch = rawArguments.match(/^["'](.+?)["']/)
+  const prompt = taskMatch?.[1] || rawArguments.split(/\s+--/)[0]?.trim() || DEFAULT_PROMPT
+
+  const maxIterationMatch = rawArguments.match(/--max-iterations=(\d+)/i)
+  const completionPromiseMatch = rawArguments.match(/--completion-promise=["']?([^"'\s]+)["']?/i)
+  const strategyMatch = rawArguments.match(/--strategy=(reset|continue)/i)
+  const strategyValue = strategyMatch?.[1]?.toLowerCase()
+
+  return {
+    prompt,
+    maxIterations: maxIterationMatch ? Number.parseInt(maxIterationMatch[1], 10) : undefined,
+    completionPromise: completionPromiseMatch?.[1],
+    strategy: strategyValue === "reset" || strategyValue === "continue" ? strategyValue : undefined,
+  }
+}