The problem is, there are categories of problems which are acceptable to find after certification, and there are categories of problems which are not acceptable to find after certification - fatigue life issues that manifest after the initial certified inspection window (the window between certification and the first deep inspection of the first inservice airframes) are acceptable, because they do not pose an undue risk to the aircraft before they can be discovered. This is because the fatigue testing of a new airframe design continues well beyond that of the certification testing, which only tests for such things as ultimate strength etc while fatigue life, inspection periods etc are done off the basis of longer term testing.
Components causing fires are in the category of things that should have been discovered during the certification period - there should be no risk from components like that for inservice aircraft, thats the point of certifying the compoments...
Out of all the problems the Boeing 787 has suffered over its so far short life, the bulk of them have not been engineering issues - only two major issues have been linked to engineering quality, and that is the side of body join problem and the initial arcing problem which caused the first airborne 787 fire during testing.
The 787s wing, designed and built by the Japanese, has proven to be better than expected spec wise.
The 787 fuselage sections built by Spirit have proven to be bang on spec.
There have been a few QA issues with the empennage and other parts, but nothing major.
The major problems stem from the decision to roll out the 787 as an essentially mocked up CFRP model on the 7/8/07 - rather than wait for the build process to proceed in the planned stages, management pushed for the aircraft to be ready for the public reveal. This lead to non-aviation-grade materials to be used to mock it up, and the aircraft had to be essentially rebuilt in the most difficult way possible afterward. This management decision made a 3 month delay into a 18 month delay.