feat(rollout): node-based rollout pipeline

[ElmoPA]

feat(rollout): node-based rollout pipeline

Rollout policy -- when to query the model, how many predicted actions to commit,
whether to blend overlapping plans -- lived as imperative code in two divergent
places: PipelineAlgo.inference_step (action queue + a PUSHSHAPES_PLAN_BLEND env
var read mid-function) and PolicyRollout.rollout_step (i %% query_frequency plus
hardcoded action slicing). Neither was configurable and they disagreed.

Training already solved this shape: an ordered list of stages over one dict,
where the list IS the control flow. Same contract here, node(state) -> state:

ObsCadence WHEN to query (on_queue_empty | every_n | always)
ObsAssemble obs -> model input (runs the training transform list)
PolicyStep algo.step, NOT forward_eval (which is teacher-forced)
ChunkCommit WHICH actions execute (n_keep, temporal-ensembling blend)
ActionDequeue one action per control step
ActionToRobot model action space -> robot interface

Defaults reproduce rollout.py exactly: QUERY_FREQUENCY=30 at 30 Hz, i.e. one
inference per second and 30 actions per plan. Verified the emitted action at
step i equals chunk row (i %% 30) of the plan made at (i//30)*30.

Nodes declare reads/writes and RolloutPipeline validates the key flow at build
time, so an ordering mistake fails at startup instead of surfacing as a None
mid-episode on hardware.

feat(robot): default rollout actions to base frame

rollout_step assumed the policy emits CAM-frame actions and undid that with the
extrinsics. That is right for a checkpoint trained through Eva.get_transform_list
(cartesian -> camframe; cartesian_wristframe_ypr -> camframe -> wrist frame), but
wrong for the fold pipeline, which applies no frame transform at all -- so the
model already predicts absolute base-frame poses. Undoing an encoding that was
never applied moves the arm smoothly to the wrong place with no error raised.

action_frame=base is now the default; cam restores the previous path for the
older checkpoints until those are refactored.

feat(robot): deploy graph policies on Python 3.10

feat(pipeline): port shared inference graph to consolidated runtime

[ElmoPA]

Warning

This pull request is not mergeable via GitHub because a downstack PR is open. Once all requirements are satisfied, merge this PR as a stack on Graphite.
Learn more

• feat(rollout): node-based rollout pipeline #588 👈 (View in Graphite)
• feat(dfot): diffusion-forcing algo, evaluators, configs and tests #587
• feat(sim): add PushShapes Sim V2 with rigid parallel-gripper physics #586
• feat(pipeline): port arc-length action tokens #585 : 1 other dependent PR (#589 )
• feat(pipeline): consolidate Fold, PushT, overlays, and MoE #579 : 2 other dependent PRs (#580 , #583 )
• main

This stack of pull requests is managed by Graphite. Learn more about stacking.