Struck-through elements covered here: T114655

For some sampled pageviews, capture:

• "title" of section
• section impressions (how many times did each section appear) - if we have 500M section opens, is that good or bad? we need to know what % were opened or closed
• page impressions - how many times was the page opened (it is much simpler if we instrument this in event logging)
• Bottom of page impression - how often does the reader reach the bottom of the page
• Article id or category (so we can look at category)
• total number of page sectoins
• total uncollapsed length of article "centimeters" - if all sections expanded = 145", this would be the number used)
• amount scrolled through (literal scrolling if possible--so if all sections collapsed = 4" and user scrolls to bottom but doesn't expand section, user scrolled 4")

The goal is to be able to say:

• the highest % of sections opened per section impression occur when the section is in position X (by how much)
• the highest % of sections opened per section impression occur when the section title is X
• X sections are opened per pageview
• X sections are closed per pageview
• X % of the page is generally seen by user (4"/145")
• X % of pageviews bring user to the bottom of the page

And "these numbers vary by category, pagelength"

Part of this analysis: https://docs.google.com/document/d/1_Efc9xog_4z_i7jkz-pikJ-j-nymV-ox8EvGa5Ae5bY/edit

Related task: https://phabricator.wikimedia.org/T114655