document user score configuration in admin guide

[cassidyjames]

The user scoring/strike system was 'coming soon' in Cove's docs, but it looks we have it fully implemented?

docs/user/concepts.md documents the formula and thresholds, but there's no guidance anywhere on how admins configure penalty weights per action in the Coop UI.

We should add a User Scores section to docs/user/administration.md (after Actions) covering how penalty weights are configured per action+policy combination, the four weight levels (Small 1pt, Medium 3pt, Large 9pt, Extreme 27pt) and when to use each, what 'strikes' means in practice for repeat violations, and a link to concepts.md#user-score for the scoring formula.

And of course, we should confirm this is exposed in the UI as I haven't actually been able to set it up and see it in action.

Cove reference (was 'coming soon' there): https://docs.getcove.com/docs/user-strike-system

[cassidyjames]

I pulled this out of the docs rework for now because I still haven't confirmed it in the UI. Here's the docs content I had for posterity:

docs/user/administration.md:

## User Scores

Coop tracks a score for each user based on their moderation history. Scores help moderators prioritize review—a low-scoring user's content can be routed to a stricter queue or automatically actioned.

Each Action has a **penalty weight** (_None_, _Low_, _Medium_, _High_, or _Severe_), and each Policy also has a penalty weight. When a decision is made, Coop adds both weights to compute the combined penalty for that (Action, Policy) pair. Higher combined penalties push the user's score down faster.

For the scoring formula, thresholds, and an example, see [User Score](concepts.md#user-score) in Basic Concepts.

docs/user/concepts.md:

## Users

### User Score

Coop tracks a score for each user based on their moderation history. Scores help moderators prioritize review—a low-scoring user's content can be routed to a stricter queue or automatically actioned.

Each Action has a penalty weight, and each Policy also has a penalty weight. When a decision is made, Coop adds both weights to compute the combined penalty for that (Action, Policy) pair. Higher combined penalties push the user's score down faster.

The User Score is calculated based on the ratio of penalties to submissions:

`weightedPenaltyRate = sum(penalties) / numSubmissions`

#### Score Thresholds

| Penalty Rate | Score     |
| :----------- | :-------- |
| ≤ 1%         | 5 (Best)  |
| ≤ 5%         | 4         |
| ≤ 10%        | 3         |
| ≤ 25%        | 2         |
| \> 25%       | 1 (Worst) |

#### How Penalties Work

- Each action and policy has a penalty weight; Coop adds both to compute the combined penalty for that (action, policy) pair:
  - None: 0 points
  - Low: 1 point
  - Medium: 3 points
  - High: 9 points
  - Severe: 27 points

#### Example

- User has 100 submissions
- 2 actions taken with "medium" penalty (3 points each) \= 6 penalty points
- Penalty rate = 6/100 = 6% → Score: 3

[cassidyjames]

It sounds like this is similar to strikes #503 where the back-end is there, but there is not a way to associate scores with actions. So this is blocked until #156 is complete.