AWS, as with many other organizations, is always under pressure to simplify things.  The fewer the tests required the simpler and easier and CHEAPER it is to get a number of people qualified. 

The main purpose is to make as sure as reasonable the percentage probability of successful weld completion without having to do NDT on every inch of every weld.  Thus, the inclusion of on position as qualified when another is welded to.  Also, certain process/electrode usage that qualifies for another. Also, groove weld qualification making one qualified to do fillets. 

The goal has been to minimize expense and redundancy.  It does not always achieve the best results.  But, are they acceptable results?  According to the appropriate code committees.. Yes.  According to many inspectors... No.  But, we don't have a loud enough voice on the committees. 

The object is to give the customer and building jurisdictions the satisfaction that even with periodic inspections the work supplied will stand up to the required loads and usage with high safety factors for the public. 

Having said all that, a customer, or even an interested fabricator/erector can test to higher standards than the code requires.  Give one more test.  So what if it takes a little time and steel.  You now have more confidence the welder can produce in exactly the joint and position in use. 

Have a Great Day,  Brent

Absolutely !

Building a quality system that exceeds code requiements is often the best practice!

The code is not the end, it's the minimum!

Testing and training welders on things that mimic what is seen in production is valuable.  Simply adhering to code for the purpose of compliance has questionable value in situations where production differs from the testing configurations called out by the code.

Edit: 
Countermeasures to an issue like the one the OP describes can be:

1)  increased testing of welders in configurations more like the ones used in production.

2)  implementing detailed work instructions that may include weld maps or bead sequencing on "job types" that have proven to be problematic.

3)  additional training for engineering, leadership and welders to support 1 & 2

4) engineering review of joints that commonly fail inspection on the shop floor.... If a V groove is preferable to a single bevel in the flat positon and can be prepared without a great increase in cost, and;  If the base metal is not too thick, a v-groove with a minimum allowable included angle that still allows access to the root may be preferable to the single bevel that is used in the horizontal position.