I'm guessing that we'll instead use the discard count as part of the tab's score, when ranking them?

I don't think there's any need to explicitly use the discard count. Rather, I'd look at using an average time-spent-discarded-before-reload as a better signal. We're putting so many heuristics into place that this isn't an apples to apples comparison. To start with I suggest we simply remove the limitation, and we can explore improvements later.

(The ML approach might completely obviate this, as it has a much better sense of which tab is important to the user *right now*.)

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/2b96359c3260886f02f88a1cada43a0b79b9a186

commit 2b96359c3260886f02f88a1cada43a0b79b9a186
Author: Chris Hamilton <chrisha@chromium.org>
Date: Thu Jul 19 14:35:44 2018

[RC] Allow repeated discards.

Allow tabs to be discarded more than once. The heuristics have changed
so much in in the last 6 months that this additional protection isn't
needed.

BUG= 794622 

Change-Id: I094af452b5a7190f229fefbc77f4571ac0601a17
Reviewed-on: https://chromium-review.googlesource.com/1142479
Commit-Queue: Chris Hamilton <chrisha@chromium.org>
Reviewed-by: Sébastien Marchand <sebmarchand@chromium.org>
Cr-Commit-Position: refs/heads/master@{#576485}
[modify] https://crrev.com/2b96359c3260886f02f88a1cada43a0b79b9a186/chrome/browser/resource_coordinator/tab_lifecycle_unit.cc
[modify] https://crrev.com/2b96359c3260886f02f88a1cada43a0b79b9a186/chrome/browser/resource_coordinator/tab_lifecycle_unit_source_unittest.cc