My thoughts: sub-maximal testing
I often wondered about these things and I've come to a few conclusions.  Testing is best suited if it predicts performance and is easily and cheaply done multiple times to gauge progress in training.  So a few years ago I decided to implement my own testing regime which worked exceptionally well for me and was directly correlated to my race performance.
Having a maximum heart rate of ~180 bpm I set a sub maximal heart rate range of 155-160 bpm (86-89%).  I chose a sub maximal value as I wanted to remove motivation/mental factors that can colour the results of a maximal test.  (Also it's not that  unpleasent that you avoid doing it.)  I then ride my rollers (fitted with a turbo fan) where I have ensured exact tire presurre of my other tests, and placed two large fans infront of me (minimizes heat load).  I do a perscribed 25 minute warmup followed by a 10 minute test where I try to ride a maximum distance while maintaining my target heart rate and a comfortable cadence (80-85).  The results are simple: the longer I can ride the better my fitness.  
As an anecdote.  Several summers ago I was feeling tired.  Lots of training, and  weekly crits and I thought I should be feeling better on the bike.  So I ran one of my many tests of that summer.  It turned out I bettered all of my previous tests.  Decided to jump in the car and headed to a not-so local road race (~ 450km away) for the weekend.  I turned out to be what I considered my best performance ever on a bike.  What I took from that was that despite a fatigued mental condition my body was performing really really well.  I think my sub-maximal test showed me that.  
Now I know the hazards of a small population statistics (n=1) but I strongly believe in well controlled and regular testing to help gain insight.  I don't think sporadic VO2/lactate/muscle biopsy testing for a keen amateur cyclist is the route.
my 2cents.