WISC: very strong in language reasoning and hands-on problem solving, only average with abstract-visual reasoning. High average working memory, average processing speed.
WRAML2: memory is largely consistent with WISC working memory results, except delayed story memory tracks verbal comprehension (which fits with how meaning allows cognition to compensate for other weaknesses). Finger windows is a visual working memory task. Design and picture memory are not like block design, firstly because they are memory tasks, not copying tasks, and secondly because pencil-paper brings in a somewhat different pool of visual-motor integration skills. Especially integration. Picture memory is entirely different from block design. If you've ever seen those old "Check and Double Check" activities in Highlights magazine, that's what they're like, except that you have to do "double check" from memory. There is a verbal-conceptual aspect to it, too, which lines up slightly with picture concepts.
CTOPP2: These are all average to low average scores, with the exception of phoneme isolation. nwr is an auditory working memory task. Rapid digit & letter naming fluency are both processing speed and retrieval fluency tasks. Elision and blending are both phoneme manipulation tasks, which is the higher end of phonological awareness tasks. A lot of high-cognitive dyslexics can do low-level PA within normal limits (or even above average), but get tripped up on the more sophisticated phoneme manipulation tasks (especially the old phoneme reversal from the first edition of the CTOPP--sadly did not survive the revision).
WIAT:
Reading: comprehension is average, most likely, as you say, being supported by context and cognition. But I wouldn't call it intact, when you compare these results to the VCI. That's a huge gap between listening and reading comprehension. Decoding skills are no better than low average. I don't see this score, but I bet accuracy would fall even further if you made her read in connected text (oral reading fluency), even with the support of context, due to having to do simultaneous decoding and comprehension.
Written language: Sentence composition is comprised of both the sentence combining and sentence building component scores. With the gap between the two, the sentence composition subtest score is of limited utility. The alphabet fluency score is about on a par with her other processing speed/retrieval fluency numbers. Here's the thing that often happens with sentence combining and sentence building, especially with this weak spelling score: sentence combining assesses your ability to construct more complex sentence structures, including causal relationships, comparisons, contrasting clauses. You can generate very simple sentences for sentence building and score well. OTOH, for sentence combining, you are provided with complete stimulus sentences, which is quite handy, if you are a child with a good grasp of complex sentence structure, but weak mechanics, as you can use cognition for the language aspect of the task, and copy the spelling from the stimulus. In sentence building, all you get is one target word for each item. You don't get any benefit from your skill with complex language, and you also don't get any help with the spelling. I'm going to guess that this is something like what happened with her. It suggests that the real hangup with written expression is mechanical (spelling, punctuation, etc.) in nature, which fits with her poor decoding. I'd be interested to see how she does with idea generation and organization.
At the moment, this looks like a more-or-less pure dyslexic situation, with poor encoding (spelling) being the flip side of decoding.
Math: and the same thing happened here, with low average basic skills, but strong reasoning. In third grade, automaticity with math facts is often an obstacle to computational sophistication. Not clear if that was the basis of the num ops relative weakness.
The recommendation for OG is, of course, the gold standard, and entirely appropriate. TBT at home should be compatible with that, not so much for multisensory reasons, but for phonological awareness/direct instruction in the phonetic system/incremental approach (bcs of apparent weaknesses in automaticity). All About Reading/All About Spelling is also a home-based system with OG roots. It's scripted, so the training/fidelity part of it is less of an issue. Relatively affordable, a plus.
