Scratching the plural out of prayer - Mt 6:5

This is a story about how difficult it can be to get the data right even before starting to ponder the original wording of a text.

These are the opening words of Mt 6:5 as in Tregelles and NA28:

Καὶ ὅταν προσεύχησθε, οὐκ ἔσεσθε ὡς οἱ ὑποκριταί,
And when you pray, you must not be as the hypocrites

The majority reading is as follows:

Καὶ ὅταν προσεύχῃ, οὐκ ἔσῃ ὥσπερ οἱ ὑποκριταί,
And when you (singular) pray, you (singular) must not be as the hypocrites

I am only interested in the two verbs, whether they are singular or plural.
In the apparatus of NA28 we learn that, among others, Codex Sinaiticus (א*) supports the singular verb, though with a minor variation, as indicated by the round brackets. In Appendix II we learn that א* actually reads προσευχη ουκ εσεσθε. That is, a singular verb προσευχη followed by the plural εσεσθε. The second corrector supports the text.

There are sorts of things wrong here. First of all, why is א* given as support for the double singular reading as, arguably (having one of each), it can be forwarded as support for the double plural reading?
Moreover, I don't think the reading of א* is accessible. Have a look at the images:

Normal light:


Striking light:



The transcribers of Sinaiticus on the Codex Sinaiticus website get it right, as usual. Here corrector Ca is made responsible for the intralinear correction of προσευχη to προσευχησθαι (itacism for προσευχησθε), the remainder of the text as visible on that line is the work of the first corrector (א^{a [or 1]}), while the erased text of א* is unreadable. As you can see on the image, a rewriting starts from the third letter onwards, where we have indications of something being scratched off the parchment before the current writing. The parchment is rough till the end of the line and even torn at the final epsilon of εσεσθε.
What can we say about the erased text of א*, except that it was erased quite efficiently? The original version had something that started with προσευχησ-. It seems clear to me (and Tischendorf) that the omicron of ουκ is created out of a lunar sigma Ϲ. But is there space on the line to have the version with the two plural verbs? I don't think so, since it would require an extra two letters to be accommodated on the line which is already wider than average. That means that if the first version had two plural verbs, the first of these (and only the first) was already corrected in scribendo, that is before the next line was written.
For once, Tischendorf is not much of a help in suggesting that perhaps the original error was made from -σθε to -σθε. Though this is possible, it means, again, that the error was caught and mended before the writing of the next line, and only after almost the whole line had been filled up. His scenario requires that in addition to messing up the original line, the scribe also messed up the correction, by forcing the first verb into a singular (even though originally he had it right).

Whatever the underlying error we should represent the testimony of Sinaiticus as follows:
א*: προσευχησ[illegible] ...
א^a: προσευχη ουκ εσεσθε ως οι
א^ca: προσευχησθαι (read προσευχησθε; remainder of the line untouched).
Interestingly, this suggest that א* may have intended to write the double plural before messing up, and that (I am slightly increasing the speculation value), whilst clearing the mess, the same scribe but now in the guise of the first corrector, messed up again.
Anyway א* is closer to the text of NA28 than to the variant in the apparatus, despite what the apparatus tells you, א^a can be read as supporting either wording but should probably figure as a separate reading, and א^{ca [or] 2]} supports indeed the reading as given in the text.