Caveat before discussing the scores: the CMMT, though somewhat popular in its era, was not an exceptionally well-constructed test, with issues in range of difficulty, reliability, and specificity. In terms of your variation in scores, if we assume that the differences are not merely an artifact of poor test-retest reliability (which is a pretty big assumption, with this test) a number of factors might have been involved:
1. Your age at testing: young children generally do not generate stable scores, both because of their inherently inconsistent testability, and because of the rate, timing, and diversity of typical development in young children. This applies to both the 1st grade and 2nd grade tests.
2. Retest effects: current practice does not allow retesting with the same instrument in under 24 months (in the era of which you speak, one year was the typical standard for retesting), because of score inflation effects. These are likely to be even more pronounced in children of above average intelligence than in others. It seems likely that you are of at least above average intelligence, which makes it plausible that the 2nd grade test score may have included some retest score inflation.
3. Given what you have described, it seems more likely to me that your actual IQ is probably somewhere in the 120s or 130s (more than adequate for being bored in a very small school with limited resources, and also more than adequate for being an NMF and one-year-early entrant to college), but that a lack of access to appropriate instruction may have had some score depression effects on your 5th grade test. I tend to think that, rather than you dumbing down to fit in (though that happens with disturbing frequency in young GT individuals, it doesn't usually affect formal individually-administered cognitive assessments), it is more likely that the absence of programming at your instructional level limited the development of those concrete intelligence skills that are measured by the test. This, btw, is similar to the effects of educational neglect and academically-impoverished home environments on socio-economically disadvantaged children.
I'd feel more confident about your IQ, if you had been tested with a stronger instrument.
4. That you were tested with a resource-costly instrument (two hours of individual testing with a graduate-(probably doctoral) level examiner) three times during elementary school indicates that your teachers knew that they were not meeting your educational needs, and repeatedly attempted to find solutions. (Apparently without much success, from your account, until high school.)
I am sorry that your formal educational experience was so distressing; I hope you have found some life satisfaction since.
