In my field, for a time, a very brief time, the time between submission and publication was a factor for people when submitting to journals. Some journals the time between submission and publication was close to two years while for other, equally reputable, journals the time between submission to publication was less than 3 months.
When people started complaining about the time to publications, the slower journals began to cheat. The outcomes of the review process used to be Accept, Revise and Resubmit, and Reject. To "speed up" the process they changed "Revise and Resubmit" into "Reject and Resubmit". The resubmission was treated as a new revision, although you got the same reviewers, so the submission clock was reset. While this did nothing to speed up the process for authors, it reduced the time between submission and publication (it also made the journal look more selective). As I said, it is cheating.
In regards to if it matters, while it would be nice for authors to have access to meaningful statistics, it is not something the publisher needs to provide. Some might argue that "time stamping", but realistically, in fields where articles are infrequently published in the first journal they are submitted to, time stamping does not exist since I have never seen a time stamp based on the first journal an article is submitted to.
One thing to be concerned about when the submission date changes is if the process has gone awry. It is possible that the resubmission is being treated as a new submission and not a "Reject and Resubmit". If that is the case, then you may get a new batch of reviewers (and editor). Unfortunately, the way the automated systems work, sometimes it is difficult to tell what is going on. In these cases, a quick message to the editor is not out of place.