One can go the individual route... the very entry level GPO can take advantage of that approach (the quality of that library is not at all in the same league as VSL, nor is it intended to be)... but for that approach to work each instrument must have its own unique samples.
Otherwise, anytime you use the same sample multiple times on the same pitch you will get phasing. The only way to avoid that in VSL would be to raise (or lower) the pitch of each individual part and then do a transposition back to concert pitch within the VI. The further the transposition or pitch stretching, the more likely the sound of the sample itself will degrade. This technique can be very important when creating a 2nd violin section, as it allows for unison parts with the 1st violin section, with the use of the same articulations since different samples are then being used (can be important with unison passages).
In short, using the individual approach is simply not a viable option. In theory, for a closer orchestral emulation, that approach would be correct. But, for strings, on the VSL sample level quality one would need roughly (assuming sections of 16; 14; 10; 8; 6) 54 unique solo strings libraries with each one featuring a different player to avoid phasing (at full cost including mutes for each $71,280), the needed computer power to run 54 separate string libraries, and particularly the time to tweak each and every part in detail. On top of that throw in the costs to produce such a project, and no real market for it...
It simply will not happen.
If you really want to, you could possibly explore creating up to about five parts (concert pitch, plus 1/2 step and whole step above and below concert pitch) as outlined above. My own thought would be that creative time would be better spent elsewhere.
