Orbit: state the fold-change contrast direction before interpreting up/down-regulation

After a DESeq2 run completed, Orbit summarized results in a way that was initially biologically misleading: genes were described as "downregulated" in a condition based on negative log2 fold-change, without stating the contrast direction. It later clarified the contrast was computed as `log2(Control/Treatment)`, so negative values actually meant *higher* expression in the treatment condition. The underlying numbers were correct -- only the interpretation was wrong because the contrast direction wasn't stated.

Orbit should explicitly display the fold-change definition (e.g. `log2(Control/Treatment)` vs `log2(Treatment/Control)`) before labeling genes up/down-regulated, so the narrative can't invert the biology.

Same scientific-correctness-guardrail family as #220. Surfaced alongside the sample-labeling issue #288.