thesis horizon

The expected duration, in MONTHS measured from the snapshot's observedAt, until this thesis resolves - either confirms (the structural call is realized) or trips its invalidationCondition. Authored as an embedded distributional estimate; TriangularEstimate is the natural family ("soonest / most-likely / latest resolution"): estimateMode = the most-likely months to resolution, estimateLower95 / estimateUpper95 = the soonest / latest envelope, estimateMean = the moment for renderers. Range is the abstract DistributionEstimate, so the embed must narrow to a concrete subclass per the embedded-type-must-narrow-range rule. Horizon is INDEPENDENT of confidence - a high-confidence thesis is not necessarily a fast one - and independent of strength: it answers only "for how long do we hold this view." It is the time anchor a downstream option-expression needs to choose an expiration; the strike comes from a separate price cone (underlying volatility x sqrt(time), drifted by conviction), not from the horizon. A target resolution DATE is derivable as observedAt + horizon, so storing a snapshot-relative duration keeps the value reusable as the snapshot rolls forward.