Create repro for mshv get dirty page issues

When using dirty page tracking on mshv in #683 a couple of issues cropped up in the mshv implementation:

1. After enabling dirty page tracking the first call to get the dirty page bitmap reports that all pages are dirty, and sometimes also includes a bit for a page beyond the end of the mapped page.
2. Calling get_dirty_log on the mshv crate gets slower as the size of the mapped memory increases, mapping 1GB of memory causes this function to take more than 1ms on some tests, this seems to be independent of the amount of data that is dirty.