Is dryer (= less room ambience including room modes and short first reflection times) always better for the final mixing result, or is there a threshold below which it simply doesn't matter, and the optimization of other dimensions, such as tonal balance, transients, harmonic purity and tonal consistency becomes more important? How do you find the best compromise between signal dryness and how good it sounds raw, without being able to hear the final result in a professional mix?
"Completely dry" gives the sound engineer complete freedom to supplant any amount of reverberant room acoustics without causing any inconsistencies, allowing free location of all musicians in a virtual room.
So it's the best material to be working with. That doesn't mean that it's an absolute prerequisite for the results to turn out good. But the fewer tracks require special care, the more likely the sound engineer has freedom to focus on producing quality.
In addition to what user90207 said, dry vocals play nicely with several common effects:
Pitch-correction software may have difficulty tracking pitch on very wet vocals, or more commonly the reverb tails just sound unnatural when a note is corrected.
If there is reverb in the recording, it will be affected by compression in a way that may not be desired. A dry vocal lets you do all of your compression and then add reverb later. If you took the amount of compression used in modern metal recordings and put that on a nice ambient recording, you'd get an explosion of reverb after each line.
Distortion also has a compressing effect. Putting a wet vocal through distortion gives a noisy track.
There is also the possibility that reverb tails in a wet vocal cause difficulty when editing takes together ("comping").
These are all standard processes in modern recording, so studios tend to record vocals very dry.
Now, to directly answer the question- It largely depends on how heavily you will use the above effects. I've had good results close miking in a mostly untreated but dead room (rug, soft-ish ceiling tiles, big bookshelves, I hung one blanket off a mic stand by the singer). In the mix, it sounds fine, even after heavy compression and pitch correction. If the vocal is especially exposed for a bit, you could just do more takes and less pitch correction. But really, my "noise floor" (measured between two syllables on the words "we could") is about -45dB, and most of that is the direct sound of me at the mic. That's plenty dry enough. Close miking does a lot for rejecting room noise.
I'm not sure what you mean by "tonal balance, transients, harmonic purity and tonal consistency". Transients can be adjusted with compression or transient shaping plugins, and everything else can only be made worse with room reflections.
I'd suggest keeping the vocal signal that you actually record as dry as you can, but put some nice effects on the vocal signal going to the singer's headphones as they perform.
That way they'll feel comfortable, but you'll have an unaffected vocal recording giving you the freedom to try other signal treatments.
It also makes punch in and comping easier because you don't have to worry about clipping reverb tails.