Lvmguider

✨ Improved

* [24](https://github.com/sdss/lvmguider/pull/24) Ensure group write permissions for new files in Docker container.

⚙️ Engineering

* Format and lint using `ruff`.
* Relax Python requirement to `>=3.10,<4`.

🔧 Fixed

* Fixed dynamically setting `additionalProperties: true` in the schama. Apparently something has changed with `jsonschema 4.22.0` and now changing the schema in the actor is not sufficient; one must then recreate the validator with the modified schema.

🔧 Fixed

* Fix comment for keyword `REFFILE` being too long when the keyword is populated.

✨ Improved

* Added `AIRMASS` and `GUIDERV` keywords to co-add FITS files and data frames.

🚀 New

* Added `RA`, `DEC`, `FWHM`, `ISFSWEEP` to the `PROC` headers, `FrameData`, and `lvmops.agcam_frame`. All AG frames are now loaded to the database, not only those associated with an exposure.
* `Focuser` now uses a temperature model to estimate the initial best focus.
* Added `adjust-focus` and `focus-info` commands. If a focus sweep has been previously executed, `adjust-focus` can be used to adjust the focuser position based on the delta temperature between the bench temperature during the focus sweep and the current temperature.

✨ Improved

* Use inverse standard deviation as weights for the focus spline fit.
* `Focuser` now has an option to require the best resulting focus to be in the range of focuser positions tested or it will automatically repeat the focus sweep with a larger step size.

✨ Improved

* Upgraded `astropy` to 6.0.0. The docker image now tries (but continues if it fails) to update the `astropy-iers-data` package before running the actor and internally `lvmguider` should never try to download IERS data over the internet.

🔧 Fixed

* Fixed `pandas` deprecation in `lvmguider.extraction`.