The issue is one that must in my opinion be viewed from two perspectives:
1) If your desire is to actively change the pH during the mash, then you must measure pH very early, while there is still saccharification taking place.
2) If your desire is to determine the most accurate reflection of mash pH overall you must measure late, sometime post saccharification and pre-boil.
The problem with the former is that mash water has not yet permeated into the deepest confines of the grains and thereby released all of the acidity (be it on a relative basis acidic or basic with respect to the mash pH target) bound within the grains, so clearly such a measurement is not going to provide a true reflection of mash pH.
The problem with the latter is that it does not allow for the time within which to chase the tail of the mash pH. But this is not really a problem, but rather is a reflection of reality.
Most grists, though not all, with this being dependent upon (among other things) the diastatic power of the grist as a whole, have essentially completed their saccharification phase by the 30 minute mark, as can be attested by iodine testing. So perhaps a valid mash pH is impossible to attain any earlier than at this juncture.
Let's look at this from the perspective of a light colored Pilsner base malt SMaSH recipe. For this case, the grist (albeit, as for all grists, acidic) is actually quite basic with respect to the mash pH target, and therefore acid has most likely been added to the mash water. At 10 minutes post doughing in the basic pH nature of all of the grains has not yet been liberated into the still actively developing wort, since saccharification is still actively in progress. So at this juncture the acids added to the mash water have not yet had time to locate and act upon fully all of the essentially basic grist, and be neutralized to some extent thereby, so an artificially low mash pH is measured due to some as yet unreacted acid still left over from the initial acid addition. A mash pH taken early for this case is clearly incorrect and biased artificially to the low side as a consequence (and as a consequence mash pH software that tends to give lower pH forecasts overall is simultaneously and equally artificially and unjustly vindicated). Given more time and permitting all of saccharification to run its full course, for this case a latter pH measurement will come in higher than an early measurement since less free acid and also lesser of some unreacted minerals are present within the mash water. And this late mash pH reading will be correct.
And for the opposite case whereby the grist for a robustly dark recipe is highly acidic with respect to the mash pH target of choice, and whereby a base such as baking soda has been added up front, the opposite case occurs. And for this case an early pH reading indicates a falsely high mash pH as a consequence.
Until saccharification has fully run its active course, which may in some cases require fully 90 minutes of mashing, a true reflection of the "overall" completion of all of the underlying reactions is not yet finished, and a static pH level has not yet been achieved within the wort.
So the best time to test for "true" mash pH may be right before the raising of the mash temperature to achieve "mash out", or for those of us who do not "mash out", wherein some small level of saccharification may persist right up to the completion of vorloft, for this case it may be best to take the mash pH sample right at the conclusion of the vorloft step.
And in all of this circulation will allow for better mash pH readings than will merely stirring the mash occasionally.
Again , this reflects merely my opinion on this matter.