[analyzer] Add one-shot and streaming BLAKE3-256 hashing

Prepare a faster replacement for the MD5 hashes used for file contents
and API signatures. Computing these hashes remains significant work even
when cached analysis avoids parsing and resolution. A pure Dart
implementation avoids native-library packaging and loading requirements.

Add ScalarBlake3 for one-shot hashing and ScalarBlake3Sink for
incremental input, finalization, and reset. Both produce full 32-byte
digests.

Add Blake3ApiSignature with a fixed 4 KiB staging buffer. Small
signatures use one-shot hashing; larger signatures stream through the
sink without accumulating the entire message. Preserve ASCII string
specialization and split values across buffer boundaries without
padding. Reading a digest finalizes the signature.

This CL adds the implementations and tests. Switching the analyzer's
existing MD5 call sites remains a follow-up.

An integration experiment used Blake3ApiSignature for API signatures and
ScalarBlake3 for file-content hashing. On an Apple M3 Max with Dart
3.15.0-edge ARM64 AOT, analyzing a frozen pkg/analyzer corpus of 1,630
Dart files (40.1 MB) produced these median times:

                         All MD5    All BLAKE3-256    Time reduction
  Populated disk cache:
    Wall                   1.604 s         1.472 s              8.2%
    Total CPU              1.635 s         1.500 s              8.3%
  Empty disk cache:
    Wall                  12.722 s        12.448 s              2.2%
    Total CPU             24.845 s        23.990 s              3.4%

Use scalar compression initially, with a planned upgrade to SIMD once
the analyzer's minimum SDK moves to Dart 3.14.

Change-Id: I7a172efd591b1da5d6871104b9287bdd55868f70
Reviewed-on: https://dart-review.googlesource.com/c/sdk/+/562480
Reviewed-by: Konstantin Shcheglov <scheglov@google.com>
Commit-Queue: Konstantin Shcheglov <scheglov@google.com>
Reviewed-by: Phil Quitslund <pquitslund@google.com>
Reviewed-by: Samuel Rawlins <srawlins@google.com>
Reviewed-by: Jens Johansen <jensj@google.com>