feat: lean per-step table w/ per-mode hack cols, generic elicit, ship->deploy

Streaming table (StepLogger) redesign per user review:
- drop sprd/N/refr from the streaming view (constant / in argv / always '-')
- short names: cos_pre->cin, cos_pre_s/t->cin_s/t, cos_post->cout, gradn->gn
- 2 sig figs on loss; 1 on gn/lr
- cin/cin_s/cin_t/cout/fired only on projecting arms (no vanilla cos_post_cf)
- ADD per-mode cumulative student-hack columns hk_<rt|eq|xc|so|se|fm> on
  multi-mode (substrate) runs -> shows WHICH loophole classes are learnt
- self-decoding legend() (only the columns this arm/mode-set shows)
- end-dump auto-renders any (n,d) tuple as n/d; drops sprd/N too

derisk_loopholes (#139): replace the 6 spoonfed exploit recipes with ONE
generic elicit (the faithful hint already discloses the mechanism; the model
must connect loophole+permission -> exploit = honest discoverability test) +
an exit-interview '### Notes' section, surfaced in the log (too-vague/too-blatant
hint signal).

Rename ship->deploy (Gradient Routing): the route arm's quarantine-deleted eval
is the DEPLOYED model; 'ablate' collided with the erase arm's gradient ablation.
train.py columns + row dict + plot_dynamics + plot_route_evidence + results.py;
dropped the dual-name back-compat reads.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

scripts/results.py

@@ -146,7 +146,7 @@ def main() -> None:
     # first-class arm now, so include it (keyed on `arm` below so it doesn't
     # merge with projected). NOTE: routing's L5_hack here is the TRAINING-time
     # hack (the routed forward still hacks); the deployment number is the
-    # ablated-eval (ROUTE EVAL BLUF / hack_abl), not this column.
+    # deploy-eval (ROUTE EVAL BLUF / hack_deploy), not this column.
     j = (df.filter(pl.col("arm").is_in(["projected", "routing"]))
            .join(van, on=["mix", "seed"], how="inner")
            .with_columns((pl.col("L5_hack") - pl.col("v_hack")).alias("dh"),