data-set commands gate on getActivePieceCount, which reverts out-of-gas on larger sets, making them uninspectable and unterminable

[TippyFlitsUK]

Version: filecoin-pin 1.3.0, mainnet, PDPVerifier 0xBADd0B92C1c71d02E7d520f64c0876538fa2557F.

data-set show, data-set ls and the terminate pre-flight all call getActivePieceCount(setId). On larger data sets this view reverts out of gas:

Error: The contract function "getActivePieceCount" reverted
vm error: 00: f03677600 (method 3844450837) -- contract reverted at 75 (33)
          01: f03677600 (method 6) -- out of gas (7)

Measured on one account's sets: 3,452 and 6,882 pieces inspect fine; 18,807 and 29,608 pieces revert. Effects:

1. data-set ls fails for the entire account as soon as one set crosses the threshold.
2. data-set show <id> fails for any such set.
3. data-set terminate <id> fails for any such set, and worse, reports a misleading error: dist/data-set/run.js wraps getDetailedDataSet in a catch that maps any error to "Could not find data set with ID X", so the revert surfaces as "not found".

Net effect: the larger a data set grows, the less able its owner is to see or terminate it through the CLI. Termination does not need a piece count at all, and the provider-relayed termination path avoids the call entirely. (This is the same view-level OOG family previously seen with getActivePieces/getActivePieceCount on million-piece sets, where cursor-based paging via getActivePiecesByCursor was the workaround - it just bites much earlier than expected, somewhere between ~7k and ~19k pieces.)

Suggested directions: drop the piece-count pre-flight from terminate, make it non-fatal in show/ls (display "unavailable" instead of failing the command), and surface the real error instead of "not found".