show latest kalman_watch plotly plot with STK radiation model

[javierggt]

Description

After kalman_watch 0.6.0 (after sot/kalman_watch/pull/16), the IR flag fraction plot will be a plotly plot, and not PNG. This PR changes the web scraper to show the new plot.

Interface impacts

Testing

Unit tests

• No unit tests

Functional tests

Not clear what is the best way to test, because everything is hardwired to the flight values.

I did the following:

from twiki_wg import ssawg_trending_scraper
class KalmanWatch3Page(ssawg_trending_scraper.KalmanWatch3Page):
    def get_url(self):
        return f"https://icxc.cfa.harvard.edu/aspect/test_review_outputs/kalman_watch/pr-16/yearly/", ""
page = KalmanWatch3Page()
page.parse_page()
chunks = page.get_html_chunks()

and checked that each of the chunks corresponds to the right element on the page.

[javierggt]

While doing this PR, I realized that the scraping can fail if something changes in plotly at some point (but hopefully it won’t). Still, it would be good to keep it in mind.

The scraping relies on the list of div elements, but that is a straight list, whereas divs come in a tree. Currently, each plotly plot contributes two div elements. If they decide to add another nested div, that will change the count and the wrong info will be displayed instead of the second plot in the kalman watch section.

[jeanconn]

I think the twiki-wg scraping code is not easily-maintainable code and we're continuing to use it knowing full well that it is brittle. I suppose I would wonder if there is a nice way to do some sanity check on the div it grabs for the plotly plot and have twiki-wg throw an error if it doesn't find what it is looking for (because in SSAWG we might miss that we're just getting an empty div or the like in review)? but that would be a twiki-wg "desirement" not required in this PR.

[jeanconn]

The noted functional testing looks good for this application.

[javierggt]

anyway, I was prompted by your comment to produce a script to check the whole thing (PR #35 ).