Interesting comments. I shall attempt to address them in depth.
Firstly, it's probably worth me providing a bit of context. As it's a studio recording, I've got about 50 mins of takes of this, one full runthrough and then multiple takes of individual sections, where anything which wasn't quite right is retaken until I feel it is. Editing then involves stitching everything together, paying specific attention to dynamic and tempo compatibility at the joins. This sometimes limits what can be used as consecutive takes, and there is one short segment here where I will need to manipulate the tempo in order to best maintain rhythmic integrity. (I, or more exactly the editing engineer, have tried to minimise the need for that - unwanted sonic artefacts can arise therefrom.) The implication of this is that I'm largely stuck with the takes I already have, and thus, short of rehiring the hall for emergency retakes - viz. those thirds :@ - I've got to do my best editwise with what exists.
Continuing -
Re D[2] I agree. I will check if there are more precise tremolandi within the takes: I've already been quite fussy, but there is certainly at least one which is impure. The thirds I definitely wasn't comfortable with (they aren't a technical forte of mine, tbh) and there is one scale which tails off one third early; that specific issue can be corrected ok.
D[9]: the thirds (again) are somewhat rhythmically nebulous, accompanimentally speaking. Excluding the segment I alluded to above they are my main gripe, unless I'm missing something.
D[19] What I would term 'abandon'. Probably the hardest, and most abstract, issue pertaining to recording. The practical aspect is that hiring good halls costs (a lot of) money, and you have only a limited time to spend on each piece. Trying to 'get it right' quickly, and simultaneously playing and self-assessing is completely antithetical to 'abandon'. Sometimes moments come along where things do just click, but I think it's easier to get such a feeling during live performance, where you're playing for the moment, not posterity, and the audience also has visual cues to potentially enhance the experience.
Thanks for taking the time to write such detailed comments.