Description for `filter_timeline_limit` config is misleading

This issue has been migrated from [#13523](https://github.com/matrix-org/synapse/issues/13523).

---

### Description

Looking at the documentation for `filter_timeline_limit`: [`docs/usage/configuration/config_documentation.md#filter_timeline_limit`](https://github.com/matrix-org/synapse/blob/c34791ef5b612d36c52216a4d9c84ee56ac00c28/docs/usage/configuration/config_documentation.md#filter_timeline_limit)

This says it would set the limit for the events returned on `/sync`. But it actually sets the upper limit for the limit parameter.

For example the default for the `filter_timeline_limit` is 100, but the actual default limit for events returned during `/sync` is 10: https://github.com/matrix-org/synapse/blob/c52abc1cfdd9e5480cdb4a03d626fe61cacc6573/synapse/api/filtering.py#L302

One way to fix that is by simply making the documentation more explicit, but I think it would actually be helpful to also rename the limit configuration parameter to something like `max_allowed_filter_timeline_limit` or so.

### Steps to reproduce

Read the docs linked above.

### Homeserver

### Synapse Version

1.65.0rc1

### Installation Method

_No response_

### Platform

### Relevant log output

```shell


```

### Anything else that would be useful to know?

_No response_

Start with docs/usage/configuration/config_documentation.md at the filter_timeline_limit section and compare its wording with synapse/api/filtering.py around line 302. Clarify that the setting limits the allowed limit parameter rather than directly setting the number of /sync events returned; decide whether the documented name should also be changed.

Written by the indexing model from the issue text.