Add split-loop-with-lambda learning hour: draft, pre-factored exercise, solution

- Samman learning hour: 5/5/10/25/5 outline; pairs stop at extracted
  method taking the lambda as a parameter
- exercise/java: pre-factored variant of the kata (Q1/Q4 dead branch
  stripped, dates as constant), approval test red/flaky by design
- exercise/solution: collectExpectedWeather(LongFunction<Boolean>) +
  no-arg overload; deterministic approval test (all-sunny lambda)
- rehearse-with-agent.md: rehearsal setup with qwen + java-refactoring-mcp

ideas/split-loop-with-lambda/exercise/java/README.md

@@ -0,0 +1,37 @@
+# IceCreamForecasting — Split Loop with Lambda (exercise)
+
+Pre-factored variant of [IceCreamForecasting-Refactoring-Kata](https://github.com/emilybache/IceCreamForecasting-Refactoring-Kata):
+the Q1/Q4 dead branch is stripped and the hardcoded dates are a constant now.
+Everything else — including the noise in the scoring loop — is untouched.
+
+## Task
+
+`longRangeForecast()` is hard to test: it calls `scorer`, and `Scorer`
+stands for an external weather service you must not call from a unit test.
+The rest of the method's inputs are only `DATES` and `ANCHOR_DATE`.
+
+Get it under test:
+
+1. First get any test running that calls `longRangeForecast()` —
+   it will fail or be flaky, because of the dependency on `Scorer`.
+2. Then use the **Split Loop with Lambda** mechanics (6 steps, on the handout)
+   so that the weather collection becomes an extracted method that takes the
+   weather-reading as a **lambda parameter**.
+3. Stop when you have an extracted method that takes the lambda as a parameter,
+   and a test that passes **without** touching `Scorer`.
+
+## Constraint
+
+Don't change `Scorer` and don't change the *behaviour* of
+`longRangeForecast()`. Compile early, compile often.
+
+## Debrief questions
+
+- At which step did plain **Split Loop** stop working, and why?
+  (Hint: what type did your introduced variable end up being?)
+- There are **two** calls to `scorer` in the weather loop.
+  What did you do with each one?
+- Could you delete either call without changing the forecast?
+  Try it. Why did that work?
+- Where would you take this next — more extraction, or moving the method
+  to another class where `Scorer` is unknown?