Sylva, welcome to the forum.
To me, as a strings player, the "perfect" strings library does not exist. Indeed all sample libraries (any instrument: strings, woods, brass, etc.) are only approximations. A sample library simply cannot truly replace a live player. A live player can be far more expressive on any given note than a corresponding recorded sample. The very fact that samples are recorded makes them static - though all sorts of controller adjustments will certainly help.
That being said, to my ears, VSL comes the closest to the actual sound of a traditional symphonic string section. To me, one of its weaknesses is in some of the release samples. The high number of articulations helps a great deal, as that better reflects what a player can actually do with a bow, compared to some libraries that have offer very few options.
If one things back to the "state of the art" 15 or 20 years ago, the overall quality of samples have come a long ways.
On a more technical level, I strongly suspect that what you are calling "hiss" and "grain" is simply a legitimate part of genuine string sound: real bow noise. Though it may be hard to believe, the type of horse hair used in a bow (for bassists it can be traditional bleached white, unbleached black, or a mix of both) has a real influence on the overall sound produced (white is by far the smoothest, black gives more grit and bite). But, even with the traditional bleached white, as used by all orchestral string sections, the rubbing of the horse hair against the strings will produce some grit. If a sound sample is missing that "grit" the sample is actually less "real". If a composer wants to emphasize the "grit" from the bow, the sound can get very harsh (as in the VSL articulations noted as such).
Also, the type of rosin used can make a difference, and both too much/too little rosin can cause problems. Depending on bow stroke used, there may also be legitimate attack noise.
Although I will not "name names" one of the lesser competitors of VSL in the past had recorded so much attack noise in all their string samples, that the net effect, IMO, was to mirror very sloppy, 2nd rate string players, and removed that library from ever being a purchase consideration.
At great distances from the player, bow noise will be less evident, and the larger the ensemble the less precise things will be (for example there is a very real difference in the string sound between the violins in chamber strings and the violins in AP).
Finally, to get anything close to a "real" sound using samples, a great deal of work is generally required. While some companies emphasize usability "out of the box" the "out of the box" libraries still require a great deal of tweaking unless you want the "stock sound". IMO, VSL's samples cover a much broader range of musical styles.