Batch-write historized time-series data to points via the Haystack hisWrite op

Steps

1. For a single point, send a grid to the hisWrite op with grid meta id set to the point's Ref and rows of ts/val pairs to append
2. For a batch write across multiple points, omit the meta id and instead define multiple value columns (v0, v1, ...), each with column meta specifying its point's id, sharing a common ts column
3. Ensure every point in one batch request shares the same configured timezone, since batch hisWrite does not perform timezone conversion
4. Match each posted value's kind and unit exactly to the point's configured kind/unit - unit conversion is explicitly disallowed by the spec
5. Expect an empty grid response on success; servers must gracefully absorb any duplicate or out-of-order samples you resend