Welcome!
There are many possible explanations for the differences in the data you've reported, so yes, having some subtest numbers might help in understanding his profile. For example, the KBIT-2 is composed of a verbal and a nonverbal component, which tend to be roughly predictive of reading/writing/language achievement and math achievement, respectively. You describe relative strengths in reading comprehension, which could mean, hypothetically, that he scored above the cutoff on the verbal portion of the KBIT, but below on the nonverbal, with a combined scored just below the cutoff. Another possibility is that a diverse learning profile wasn't fully captured on the KBIT, as it is considered a brief screening instrument, not a comprehensive measure. Similarly, the KTEA has many different components, some of which are more related to efficiency (e.g., timed fluency tasks) than to high level skills or applications. 95th %ile in everything would exclude quite a few deep, deliberative thinkers, as well as those with extreme focal strengths.
But even if we take the available data at face value, it's also important to recognize that students who just miss the cutoff for formal GT programs are still above average, and may have learning needs for advancement or enrichment not provided in the standard track. From that angle, the most important assessment data to consider are 1) his response to being placed two years up in math, 2) his request to compact 7th and 8th grade through homeschooling, and 3) your observation that the current environment is not effective. It sounds like he and you both experienced homeschooling positively (or he wouldn't be requesting it); if HS'ing is compatible with your family situation next year, that would appear to be a reasonable direction to explore.
