Update read.bedMethyl() and alpha version of implementation for working with 5hmC and 5mC in bsseq()

[SrenBlikdal]

Summary of Updates

This PR introduces several improvements related to handling long-read data imported as MethylCounts and integrating it more smoothly into the BSseq workflow.

1. Validation of Long‑Read MethylCounts Input

• A new check added to "guess" if input is suitable for likelihood filtering, and
• to detect which modifications are present in the input files.

This helps and can be extended to ensure that downstream processing is applied appropriately.

2. Updates to the BSseq Constructor
   The BSseq constructor now supports MethylCounts objects directly.

• A new options for handling 5hmC, and
• A more explicit control over what is included in the M matrix.

3. Improvements to Likelihood Filtering Code

Minor readability and clarity improvements have been made to the likelihood filtering workflow.

Notes

Some of these updates may require corresponding changes to the BSseq documentation if we decide that this will be the final structure. At this stage, this should be considered an alpha version.

If you think it would be better to place this work in a separate branch than devel for now, please feel free to do so.

I have added a user guide describing the contents of this PR (PDF linked below). If helpful, I can also add the example data to an ExperimentHub package.

User guide.pdf