You won't have any luck with bowtie2 by itself. Anyone you saw using that alone simply had no idea what they were doing and got completely crap results. You might have luck with integrating bwa-meth with galaxy, since it's pretty minimal on its requirements (just bwa I think). The better question is, "why bother with Galaxy"? It's easier (and vastly faster) to not use it than to integrate a new tool into it. If all you have is windows, then try to steal a labmate's Mac.

Regarding calling methylation peaks, you don't want to do that. Perhaps you've read papers about using within-sample methylation changes to find accessibility changes, such as promoter sites. In general, what you do after you have the resulting methylation metrics will be dependent on the biological question.

Whatever you do, make sure to make an M-bias plot before extracting methylation metrics. There's no sense in starting a downstream analysis with noisier data than needed.

If you are interested in building pipeline for bisulfite sequencing data analysis, you can do it easily on Genestack platform. What you need is just to sign up for free, upload your data and build your pipeline. It can include preprocessing of you raw reads, mapping step (Genestack Bisulfite Sequencing Mapping app is based on bsmap) and Methylation Ratio Analysis (to determine methylation percent).

Here is WGBS pipeline on Github, you may try it.

https://github.com/Czh3/NGSTools/blob/master/BS_seq.py