[RFC 60] Renumber to match PR; reframe as standalone (not rewrite)

RFC 33 was never accepted, so this RFC takes the number of its own PR (60)
and stands on its own against the current state of develop. The previous
abandoned draft on PR #51 is not part of this RFC's history.

Changes:
- Rename rfcs/0033-block-turboquant.md → rfcs/0060-block-turboquant.md
  (RFC number must match PR number per the rfcs repo convention).
- Update frontmatter to the rfcs template style (Start Date, Authors,
  RFC PR link). New Start Date is 2026-05-12; the rewrite-history line
  is dropped since there is no prior accepted RFC to rewrite.
- Strip "this rewrite supersedes the original RFC 33" framing. The
  Summary now anchors the design to the current vortex-turboquant code
  on develop and the algorithm-family prior art (TurboQuant, EDEN,
  DRIVE), without invoking the abandoned draft.
- Rename "Resolved during this rewrite" → "Resolved during initial
  drafting."

Signed-off-by: Will Manning <will@willmanning.io>
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Signed-off-by: Will Manning <will@willmanning.io>

Author: lwwmanning

rfcs/0033-block-turboquant.md rfcs/0060-block-turboquant.mdrfcs/0033-block-turboquant.md renamed to rfcs/0060-block-turboquant.md

@@ -1,8 +1,8 @@
 # Block-Decomposed TurboQuant: a Lossy Extension Type with PDX Layout
 
-**Authors:** @lwwmanning, @connortsui20
-**Status:** Proposal
-**Date:** 2026-04-02 (rewritten 2026-05-12; citations corrected 2026-05-12)
+- Start Date: 2026-05-12
+- Authors: @lwwmanning, @connortsui20
+- RFC PR: [vortex-data/rfcs#60](https://github.com/vortex-data/rfcs/pull/60)
 
 ## Summary
 
@@ -35,17 +35,16 @@ decision below:
   recompression by the default cascade compressor, and no "approximate" flag —
   the encoding is lossy by definition.
 
-This rewrite supersedes the original RFC 33 draft, which was written against an
-earlier monolithic `TurboQuantArray` design that has since been replaced by the
-extension-type model on `vortex-data/vortex` `develop` (PR #7829, merged
-2026-05-07). It also incorporates a piece of prior art the original RFC missed:
-the **EDEN paper** [15] ([arXiv:2108.08842], ICML 2022) and its predecessor
-**DRIVE** [16] ([arXiv:2105.08339], NeurIPS 2021) predate TurboQuant [1]
-([arXiv:2504.19874]). A recent note by the EDEN authors [14]
-([arXiv:2604.18555], April 2026) demonstrates that TurboQuant is a suboptimal
-special case of the same RHT + Lloyd-Max scalar quantizer family. We adopt
-EDEN's contributions while keeping the codebase's TurboQuant branding;
-see §4 "Naming."
+This RFC is anchored to the current state of the `vortex-data/vortex`
+`develop` branch — specifically the extension-type implementation in the
+standalone `vortex-turboquant` crate (PR #7829, merged 2026-05-07). The
+algorithm sits in a family that includes **EDEN** [15] ([arXiv:2108.08842],
+ICML 2022) and its predecessor **DRIVE** [16] ([arXiv:2105.08339],
+NeurIPS 2021); both predate TurboQuant [1] ([arXiv:2504.19874]). A recent
+note by the EDEN authors [14] ([arXiv:2604.18555], April 2026) shows that
+TurboQuant is a special case of EDEN with a suboptimal fixed scale `S = 1`.
+We adopt EDEN's contributions while keeping the codebase's TurboQuant
+branding; see §4 "Naming."
 
 [arXiv:2108.08842]: https://arxiv.org/abs/2108.08842
 [arXiv:2105.08339]: https://arxiv.org/abs/2105.08339
@@ -210,7 +209,7 @@ correlation tests.
 
 ### Relationship to EDEN and DRIVE
 
-The original RFC 33 missed an important piece of prior art. **EDEN** [15]
+An important piece of prior art for this design family is **EDEN** [15]
 (Vargaftik et al., ICML 2022; [arXiv:2108.08842]) and its predecessor
 **DRIVE** [16] (NeurIPS 2021; [arXiv:2105.08339]) use the same building blocks
 as TurboQuant — Randomized Hadamard Transform plus Lloyd–Max scalar
@@ -1304,8 +1303,8 @@ with the column's metadata.
 SORF at B dimensions (heuristic; real cost is dominated by memory
 bandwidth and constant factors): `R · B · log₂(B)` butterflies + `R · B`
 sign applications per block. The per-vector normalization multiplies
-(`B` per block) are omitted from these counts as in the original RFC 33;
-they amortize against memory bandwidth in practice. For Stage 1 with
+(`B` per block) are omitted from these counts; they amortize against
+memory bandwidth in practice. For Stage 1 with
 padded_dim and 3 rounds:
 
 | padded_dim | SORF FLOPs                  | k   | Total per-vector FLOPs |
@@ -2258,11 +2257,11 @@ work, not blockers on the design:
    d < 128 by a new writer would be rejected by an old reader. Surface
    this in the migration plan when the experiment lands.
 
-### Resolved during this rewrite
+### Resolved during initial drafting
 
-The following questions were live during the first draft of the rewrite
-and have been resolved here. Recording the resolutions so future readers
-don't re-litigate.
+The following questions were considered during the design and have been
+resolved here. Recording the resolutions so future readers don't
+re-litigate.
 
 - **Norm storage shape uniformity (Stage 2)** — resolved in favor of
   `Primitive<element_ptype>` when `num_blocks == 1` (matches Stage 1