At this point, I just stopped citing any article paywalled by elsevier, and only their arxiv counterpart.
If anyone ask, I will argue that I cannot obtain the version published by elsevier and have only read the arxiv version; hence citing the elsevier version would be disingenuous.
This might make a good flowchart. The decision points are:
One of the outcomes is "do not cite the paper".
Just leaving this here for everyone! This browser extension saved me a lot of hassle so many times. We need to stop paying where possible and this is the convenient legal way for a lot of papers out there: https://unpaywall.org/