You are basing a hypotheses solely on modern distribution.
No, it is an analysis involving:
--> Diversity of modern clades
--> Diversification dynamics
--> Known archeological-documented events
You are basing all your conclusion on undersampled ancient DNA (which is in fact no different than modern DNA, just with a little bit less convolution, but a very-very bad coverage).
according to you I should call my maternal grandfather who is J2b-Z631 Arthur the Celt descendant or something
Ok, I understand your issue now ... as it is "touching" your own history, you are getting emotional !
Then, you obviously confuse Haplogroups with Cultures, they are not the same.
Haplogroups carriers can exit, enter, or change of culture with time ...
If your grandfather is a Celt, then you are a time-traveller (nice to meet you, from when did you arrived ?).
From what I understand, your grandfather is an Albanian. As you ! Spoilers: none of you are "Illyrians" (some of your ancestors likey were ... but you also likely have a vast diversity of origin if we go 3000 ybp).
Y-DNA is not a culture and Y-DNA didn't summerize all your origins ! Your paternal line is no more important (biologicaly) than any other branch. Then, if you give more importance to your paternal line, it is something between you and you !
To be direct, Z631 likely formed around Slovenia/Austria, yes. The current evidence from both ancient DNA and diversification dynamic are not letting much space at that point. If you want to start Z631 in the southern Balkans, some have to migrate north very early, very fast, and way before roman times (if you provide a working mechanism, we can discuss it).
For instance, you can check a very simple statistical criteria : "diversification correlation".
-When one clade is expanding nearby other, the nearby clades will experience "small" diversification events (induced by population movements caused by the expanding population).
-This is not true for all nearby-clades, just a fraction of them will face such event. Then, to check this you need a large set of clades (what you have with Z631 by roman time for exemple).
-For instance, not seing BA-collapse for Z631 is no big-deal (even if they were likely in an affected area), not seing BA-collapse for all Z597+ branch would be an issue.
-Not seing roman impact on one Z631 would be no big deal, not seing it in all of them ... is a very-very-very big issue considering the number of clades.
For instance, look at the clade for which I think we agree:
-Y15058 : very significant diversification around roman times (this this the type of effect you should see on Z631 if it was affected by the roman conquest and incorporated in Roman army).
-Y21045 : diversification around roman and Slavic times. You can read the history of Albania in the diversification curve of this clade.
Z638 is no different with Y15058, when Y15058 diversifies most Z638 clades are having smal diversification events. This is consistent with what you would expect for nearby clades.
I don't go in details but you see such effect for all haplogroups : it work with R1b, with R1a, with N, with O ... it is really a well known thing : historical-events translate in diversification events.
After, there is people not understanding this probe (because it require some technical skills), or people not liking what it shows that will go wild about it.
But empiracal cases where this probe works is just amazingly good. Of course, interpretation can always be challenged, but dismissing this probe is not serious.
Then, let speak of Z631: nothing happen around roman times ... really nothing ! It didn't work with an Illyrian setting (diversity in Albania is not even that strong considering the level of over-sampling of this country ... Z631 is all over Europe, there is no reasons to believe it didn't went to Albania).
When is Z631 expanding ? Around 900-800 BCE, did we see something in southern Balkanic clades ? No, nothing again, should-we believe that Z631 expanded without affecting nearby populations ?
There is some Y15058 clades that could show some 900 BCE signal but very faint. It is a clade, BY162321, that is identified in 300 BCE at the Hungarian/Austrian border ... not a surprise !
Thus basically, all nearby clades have "similar" diversification histories ... we observe that everywhere, but for Z631 we should abandon this probe because in a given country some peoples are unhappy about the conclusions ?
On top of that, the oldest Z631 is found in Italy by 100 CE with a typical Gaul-like admixture.
This sample is really interresting, because it didn't carry the typical admixture of the location in which he is found.
Therefore, this sample is neither carrying the admixture from its context, neither the admixture for your theoritical origin of all Z631.
And, with this sample existing you are mocking people considering Celtic origin for the haplogroup of this 100 CE man ?
Please, try to be serious.
But anyway, some other Z631 was in the Balkans by that time (and likely before romans ...).
There is some balkanic clump with 600 BCE diversity, showing that they would not have arrived significantly later than ~500 BCE.
My current diffusion models explain all ancient samples + statistical consideration about diversification dynamic.
Then, if you don't like the conclusion, it is not really my concern ... but for sure, the one doing 2011-like work, is maybe the person that is basing all his conclusions on an isolated cousins ~800 years post-split and his strong whish of some ancestral "Illyrian" lineage.