Forecast what Pipedrive deals will close, and spot what slipped

**Objective:** Forecast the open deals expected to close in the next `{days-ahead}` days (default 90), optionally only for `{owner-name}` and the `{pipeline-name}` pipeline, and show which deals have slipped past their close date.

**Step 1: Resolve the filters**

* If `{owner-name}` is given, call `pipedrive-get-user` with `name.contains`. If more than one user matches, list them and stop. Keep the user-id.
* Call `pipedrive-get-stage` (with `pipeline-name` if `{pipeline-name}` is given) and keep each stage's name and deal-probability, to weight deals that have no probability of their own.

**Step 2: Pull the forecast window**

Call `pipedrive-get-deal` with `status` = open, `expected-close-date.from` = today, `expected-close-date.to` = today plus the window, and `owner-id` if resolved. If `{pipeline-name}` is given, keep only rows whose pipeline matches it.

**Step 3: Pull the slipped deals**

Call `pipedrive-get-deal` with `status` = open and `expected-close-date.to` = yesterday (same owner and pipeline filters). These are open deals whose close date has already passed. Keep them out of the forecast totals.

**Step 4: Report**

1. **By month:** month, number of deals, total value, weighted value (value x probability / 100), per currency.
2. **By owner:** the same columns per owner (skip when `{owner-name}` is given).
3. **Commit vs best case:** commit = deals at 70% probability or higher; best case = everything in the window.
4. **Top 10 deals** by weighted value: deal, organization, owner, stage, value, probability, expected close.
5. **Slipped deals:** deal, owner, stage, value, original expected close, days overdue.

Finish with a one-paragraph read of the forecast: how concentrated it is, and how much of it depends on the slipped deals being re-dated.

**Rules & Guidelines:**

* Read-only. Do not call `pipedrive-update-deal` to re-date slipped deals or `pipedrive-add-activity` to chase them.
* Never add values across currencies.
* Use the deal's probability when set, otherwise its stage's deal-probability. Say how many deals used the stage fallback.
* If a filter on `owner-id` returns nothing when you expect records, run the same read without it and keep the rows whose owner-name matches the user instead.

**Validation:**

* If nothing is expected to close in the window, say so and still report the slipped deals.
* If `{days-ahead}` is not a number, use 90 and say so.