Came across this while `--touch`ing many files, but it seems having
something like `max-jobs-per-timespan: "100/1m"` in a profile will also
limit the rate for files getting touched (which is a very different
operation to actually submitting jobs).

Previously you would get many messages for
> Job rate limit reached, waiting for free slots.

Whereas now you should only encounter that message if you are not in
dryrun or touch mode.

### QC
<!-- Make sure that you can tick the boxes below. -->

* [x] The PR contains a test case for the changes or the changes are
already covered by an existing test case.
* [x] The documentation (`docs/`) is updated to reflect the changes or
this is not necessary (e.g. if the change does neither modify the
language nor the behavior or functionalities of Snakemake).


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Bug Fixes**
* Touch mode now bypasses job submission rate limiting, enabling faster
execution when only updating timestamps.
* Users with a configured submission cap will no longer experience
artificial delays in touch runs; normal runs continue to honor rate
limits, and dry-run behavior is unchanged.
* Scheduling order and resource checks remain the same; only the rate
limiting is skipped in touch mode, improving responsiveness without
altering other execution semantics.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

<!--Add a description of your PR here-->

On some HPC, `module load` prints to `stdout` when loading the
requirements of the environmental modules:
```
$ module load bowtie2/2.5.4 
Loading bowtie2/2.5.4
  Loading requirement: binutils/2.40 gcc/13.2.0
```

Not sure why, but this triggers an error when `snakemake` tries to parse
it:
```
Unable to determine Python version from output 'Loading bowtie2/2.5.4
  Loading requirement: binutils/2.40 gcc/13.2.0
[3, 13]
': Expecting value: line 1 column 1 (char 0)
```

Changes have been tested on both `lmod` and `envmod`.

### QC
<!-- Make sure that you can tick the boxes below. -->

* [x] The PR contains a test case for the changes or the changes are
already covered by an existing test case.
* [x] The documentation (`docs/`) is updated to reflect the changes or
this is not necessary (e.g. if the change does neither modify the
language nor the behavior or functionalities of Snakemake).


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **Refactor**
* Updated environment module loading to suppress output during module
load operations for a cleaner user experience.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

…put (#3667)

I noticed an edge case where unused temporary output of different rules
would go to the same file. This can introduce nasty side effects, and
also made it so that my rules needed to be re-run all the times (because
two rules that produce the same output is always detected as a code
change by one of the two rules). This is now nicely raises an exception
giving an early warning to the user that sth is misconfigured.

I'm not sure how this affects performance or other stuff, let me know if
sth needs changes or if this is actually wanted behaviour 😄

)

<!--Add a description of your PR here-->

### QC
<!-- Make sure that you can tick the boxes below. -->

* [x] The PR contains a test case for the changes or the changes are
already covered by an existing test case.
* [x] The documentation (`docs/`) is updated to reflect the changes or
this is not necessary (e.g. if the change does neither modify the
language nor the behavior or functionalities of Snakemake).


<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->
## Summary by CodeRabbit

* **New Features**
* Plugin-based scheduler system with selectable scheduler plugins; new
greedy and MILP schedulers and settings exposed.
  * Exposed total temporary-file size reporting for inputs.

* **Improvements**
* CLI now picks schedulers from available plugins; legacy solver options
removed/deprecated and plugin settings integrated.
* Asynchronous input-size prefetching and stricter per-job validation
for scheduling; scheduler settings passed through execution API.

* **Bug Fixes**
  * Fixed Python version error message formatting.

* **Refactor**
* Scheduling flow reworked to use external scheduler plugins with a
greedy fallback.

* **Chores**
  * Updated help text and deprecation notices.
<!-- end of auto-generated comment: release notes by coderabbit.ai -->

🤖 I have created a release *beep* *boop*
---


##
[9.10.0](v9.9.0...v9.10.0)
(2025-08-19)


### Features

* migrate to scheduler plugin interface and scheduler plugins
([#3676](#3676))
([26fcd38](26fcd38))


### Bug Fixes

* don't rate limit jobs when touching
([#3699](#3699))
([9c499e5](9c499e5))
* raise an error when different rules produce identical (temp) output
([#3667](#3667))
([f627176](f627176))
* silence messages on module load
([#3688](#3688))
([b13e9c8](b13e9c8))

---
This PR was generated with [Release
Please](https://github.com/googleapis/release-please). See
[documentation](https://github.com/googleapis/release-please#release-please).