...All holes or only square ones? Or only over a certain size? Or something else? I'm thinking that no one seems to worry about bottle cage mount holes or those tiny little vent holes in fork legs and BB shells.
the SCF of x3 arises from a consideration of the 2D in-plane stresses around a circular hole in a (infinite) flat plate. It holds for all diameters of round holes (the amount of diverted stress balances against the effect of the hole radius on concentrating that stress). It is also close to reality for holes of a reasonable size in a thin-walled tube. However there are complications such as
a) when the hole gets smaller and smaller diameter the stresses get 'more three dimensional' and whilst there isn't a sharp cutoff, holes smaller in diameter than the wall thickness of the tube are probably a special case.
b) as previously mentioned the edge radiuses of the hole can dramatically affect the likelihood of cracking, potentially making the SCF in the region of x10 or more rather than x3
c) as others have mentioned proper bottle braze ons are reinforced. However ones done with Nutserts are not reinforced in the same way and these are by comparison a bit of a lash-up.
d) oval holes in main frame tubes are always about as bad (or worse) vs a circular hole of diameter equal to the long axis of the oval. The reason for this is that all the main tubes (but especially the down tube) are loaded in a mixture of torsion and bending; there is no 'favourable direction' for an oval hole to be oriented.
e) The primary loadings on stays are uniaxial and bending, and on fork blades bending. Breather holes in stays and fork blades are placed sufficiently far from the brazed joints that they don't see the highest bending loads. In the case of fork blades they are (over most of their length) made in such a way as they are about constant weight per unit length. Unsurprisingly they are quite thick-walled near the smaller diameter ends and here the bending loads are lowest. I don't remember ever seeing a crack at a breather hole in a steel frame, although I have seen a few that have been implicated in corrosion. Part of the reason they see low stresses is that they are commonly drilled orthogonal to the primary bending stress, so the breather hole lies on the neutral axis of the tube anyway.
f) if the hole diameter gets to be a significant fraction of the tube diameter the stresses will start to become larger than the simple SCF might predict.
g) there is usually plenty of meat that can come out of a steel BB shell quite safely; structurally speaking even quite large cutouts in the bottom of the BB shell are unlikely to be problematic, except for the simple reason that they let water into the frame. It doesn't look very 'trick' but a BB shell can be made a lot lighter in weight and still be weatherproof if cutouts are plated over with much thinner material.
cheers