There's only a limited amount of science in determining the demonstrated crosswind limit, here are the main bits of CS.23 - the European light aircraft standards. Other standards are pretty similar.
CS 23.745 Nose/tailwheel
steering
(a) If nose/tailwheel steering is installed, it must be demonstrated that its use does not require exceptional pilot skill during takeoff and landing, in crosswinds and in the event of an engine failure or its use must be limited to low speed manoeuvring.
(b) Movement of the pilots steering control must not interfere with correct retraction or extension of the landing gear.
CS 23.1585 Operating procedures
(a) For all aeroplanes, information concerning normal, abnormal (if applicable) and emergency procedures and other pertinent information necessary for safe operation and the achievement of the scheduled performance must be
furnished, including –
(1) An explanation of significant or unusual flight or ground handling
characteristics;
(2) The maximum demonstrated values of crosswind for take-off and landing and procedures and information pertinent to operations in crosswinds;
(And some of the interpretative material later on)
(3) Approach and Landing. The steady gliding approach, the pilot skill, the conditions, the vertical accelerations, and the aeroplane actions in 23.75(a), (b), and (c) are concerned primarily with not requiring particularly skilful or abrupt manoeuvres after passing the 15 m (50ft) point. The phrase ‘steady gliding approach,’ taken in its strictest sense, means power off. However, it has generally been considered that some power may be used during a steady gliding approach to maintain at least 1.3 VS1 control sink rate on final approach. For those aeroplanes using power during approach, power may be decreased after passing the 15 m (50ft) point and there should be no nose depression by use of the longitudinal control. For those aeroplanes approaching with power off, the longitudinal control may be used as necessary to maintain a safe speed for flare. In both cases, there should be no change in configuration and power should not be increased. The landing distance and the procedure specified in the AFM are then based on the power used for the demonstration. The power used and the technique used to achieve the landing distances should be clearly stated in the AFM. This applies to portions of the approach prior to and after the 15 m (50ft) height. The aeroplane should be satisfactorily controllable when landing under the most unfavourable conditions to be encountered in service, including cross winds, wet runway surfaces and with one engine inoperative. Demonstration of landing with an adverse crosswind of at least 0.2 VS0 will be acceptable and operation on wet (but not contaminated) runway surfaces may be simulated by disconnecting osewheel steering. The effect of weight
What does all this mean in practice. Basically the test team - ideally with a well instrumented runway (anemometers close to the threshold are nice) will go up, start into wind, and slowly increase the crosswind, and in some cases tailwind components by repositioning and picking their conditions.
Eventually one of two things will happen:-
(1) You've met or exceeded the 0.2Vso value, AND you can't find anything worse - so you declare the demonstrated crosswind limit at the worse you've seen.
(2) The team actually reach conditions which they consider require "exceptional piloting skill", so they wind it back by a few knots and declare a demonstrated crosswind limit about there.
Defining "exceptional piloting skill" is tough - but as a working rule HQR 5 or above on the
Cooper Harper scale would be a good basis.
One thing is also worth adding - the comment that you can't expect to fly the aeroplane better than the test pilot, is almost certainly untrue. A test pilot is a man or woman with many professional tasks - not just flying, so they are not necessarily as sharp in the cockpit as a full time professional pilot who does nothing but fly. Also, at the end of a light aircraft certification programme the most experienced test pilot on type maybe has 200 hours on type? Give it a few years and there will be working pilots with 1000+ on type, and who fly nothing else. So, it would be a bit worrying if there weren't, eventually, pilots who can fly it better than the test pilot did.
I've certainly safely landed aeroplanes that I had low 3-figure hours on type, significantly beyond the "demonstrated crosswind limit" in the book, and I'll bet plenty of other people here have too.
G