Just addressing one of your points (although several of them require critique, this quote demonstrates confirmation bias.
The data showed ECA near total resections trended better than the treatment arm. — Exwannabe
The hazard ratio was 1.01, that’s not trending, that’s as close as you can get on only one subgroup (you chose the least persuasive subgroup), and the curve itself seems to demonstrate a cross and tail, which would negate your argument that the surgeries were equal, instead, assuming no harm (as you assume water), the graph appears to demonstrate less complete surgery, on average, in the treatment subgroup which DCVax-l still manages to convert to a better long term survival.
A logical conclusion is that with complete resections, the results still demonstrate surgeons play a critical role, but it also shows that for long time survival, DCVax-l seems to provide some help, even in this subgroup, which is the least persuasive subgroup.
The remaining subgroups are more persuasive. For the incomplete surgeries, the results are stellar, and the phenomenon I label “surgeon’s helper” seems to be demonstrable. (As this phenomenon even occurred in the Celldex trial)