If the min figure is 100, and you are using the transistor as a common emitter switch, then make sure that sufficient base current is available to saturate the transistor - e.g. to switch 100mA, make sure you have maybe 5mA of base current available - this way you are sure that the transistor is 'fully switched' regardless of the part's actual gain.
No. The datasheets for most little transistors show the transistor saturating pretty well when the base current is 1/10th the collector current, not 1/20th, even if the current gain is very high.
Not much is gained by oversaturating the transistor. There are some very real downsides to the 1/10 rule of thumb. As is the case for any "rule of thumb". One must not use it mindlessly.
The 10% loss of efficiency (not necessarily so, but true in a lot of circuits) is a very real downside. The very long storage time is another. Low-Vce(sat) types have quite high hFE, even at large Ic and small Vce, and are typically spec'd for saturation at hFE = 50 or 80 (where switching time is acceptable).
Of course, if you must handle a wide range of load currents, base current must be fixed by the maximum collector load current, regardless of load*. In that case, the transistor will almost always be very oversaturated, and that's just something you have to accept.
*Unless you use a saturation control circuit; a Darlington transistor is an example, though a somewhat costly one, the cost being much higher Vce(sat). There are other solutions that perform better, but they're mostly relevant to IC design.
Tim