Basically with inadequate bracing on any manifold, you are allowing the metal to be experienced to a considerable amount of stress, both cyclic and uni-directional. For long term service life, problems such as fatigue cracking and creep damage will likely occur (also alloy-dependant).
You mentioned the brittleness of 321 as compared to non-stainless alloy steel. As a matter of fact, stainless steels such as 321 and 304 do not perform any worse than their alloy counterparts (they aren't actually hardened, they have an austenitic structure which is responsible for their mechanical properties); A P12 steel (1 Cr - 0.5 Mo) has a UTS of 415 MPa, a yield strength of 220 MPa and a min. elongation of 30% (in 50mm) Comparitavely, a TP321H (18Cr - 10Ni - ~0.5Ti) stainless tube has a UTS of 515 MPa, a yield strength of 205 MPa, and a min. elongation of 35%. These materials have quite similar properties, with the 321 having considerably better high temperature characteristics.
As for tube thickness, there is no doubt that a maximum thickness should be adhered to - if it is in fact the tube itself that fails, the tube is either a) completely unacceptable metallurgically speaking, or b) the wrong tube for the application. With quality tubes, overmatched welds, and adequate bracing, there is no reason for a manifold to fail, no matter what steel it is made of (within reason that is)
All of the tubing I am referring to is from ASTM standards A213 and A335 (Standard specs for Seamless boiler, superheater and heat exchanger tubes and Standard spec for ferritic alloy steel pipe for high temp service, respectively)
If anyone has a cracked manifold which they have junked, I wouldn't mind having a look at some of the sections to get an idea of what caused the failure.