Next steps for `fgfa window-depth`

Here are a few ways we can continue to improve the `fgfa window-depth` command, which recently landed in #263:

* [ ] The `weighted_depths` function could probably return an iterator to avoid materializing a full `Vec`. #265
* [ ] It may be possible to simplify the core `assign_depths` algorithm by turning the inner `for` loop into a `while` loop, eliminating the `i` variable, and rethinking `overlap_flag`. (This would not be an optimization; it would just be for readability.) #265
* [ ] We can try harder to match the exact output from odgi. We currently differ a little bit, ether in the calculations themselves or the printing.

Start with the `fgfa window-depth` implementation introduced in issue #263 and review the `weighted_depths` and `assign_depths` work described here. Compare the command's output with odgi, then confirm which listed improvement is being pursued and that its behavior or output matches the agreed expectation.

Written by the indexing model from the issue text.