perf(ext/streams): optimize streams (#20649)

This PR introduces several optimizations to streams

### Highlights:
- `ReadableStream` constructor: +20% iter/s.
- `WritableStream` constructor: +50% iter/s.
- `TransformStream` constructor: +30% iter/s.
- `ReadableStream` iterator (both 2 and 20 chunks): +42% and +25%
iter/s.
- `ReadableByteStream` iterator (both 2 and 20 chunks): +39% and +20%
iter/s.

### Benchmarks
**main**
```
cpu: 13th Gen Intel(R) Core(TM) i9-13900H
runtime: deno 1.37.0 (x86_64-unknown-linux-gnu)

benchmark                                      time (avg)        iter/s             (min … max)       p75       p99      p995
----------------------------------------------------------------------------------------------- -----------------------------
ReadableStream constructor                    294.52 ns/iter   3,395,392.9 (277.92 ns … 618.26 ns) 292.66 ns 353.87 ns 618.26 ns
WritableStream constructor                    235.51 ns/iter   4,246,065.3 (213.04 ns … 306.35 ns) 236.77 ns 279.08 ns 281.32 ns
TransformStream constructor                   672.52 ns/iter   1,486,938.7 (652.15 ns … 880.74 ns) 670.11 ns 880.74 ns 880.74 ns
ReadableStream - iterator (2 chunks)           10.44 µs/iter      95,757.9   (8.97 µs … 830.91 µs)  10.22 µs  14.74 µs  18.93 µs
ReadableStream - iterator (20 chunks)          21.93 µs/iter      45,593.4   (18.8 µs … 864.97 µs)  20.57 µs  57.15 µs 137.16 µs
ReadableStream - reader (2 chunks)              7.09 µs/iter     140,987.2     (7.03 µs … 7.18 µs)   7.13 µs   7.18 µs   7.18 µs
ReadableStream - reader (20 chunks)            18.41 µs/iter      54,324.2    (15.7 µs … 252.7 µs)  17.14 µs  68.88 µs  94.08 µs
ReadableByteStream - iterator (2 chunks)       11.06 µs/iter      90,375.1   (9.75 µs … 404.69 µs)  10.88 µs   16.6 µs  29.69 µs
ReadableByteStream - iterator (20 chunks)      26.71 µs/iter      37,435.0  (22.98 µs … 508.34 µs)  25.25 µs  85.28 µs 155.65 µs
ReadableByteStream - reader (2 chunks)          7.99 µs/iter     125,131.1     (7.92 µs … 8.13 µs)   8.01 µs   8.13 µs   8.13 µs
ReadableByteStream - reader (20 chunks)        23.46 µs/iter      42,618.5  (20.28 µs … 414.66 µs)  21.94 µs  90.52 µs 147.38 µs
```

**this PR**
```
cpu: 13th Gen Intel(R) Core(TM) i9-13900H
runtime: deno 1.37.0 (x86_64-unknown-linux-gnu)

benchmark                                      time (avg)        iter/s             (min … max)       p75       p99      p995
----------------------------------------------------------------------------------------------- -----------------------------
ReadableStream constructor                    235.48 ns/iter   4,246,584.3 (223.12 ns … 504.65 ns)  234.3 ns 290.84 ns 311.12 ns
WritableStream constructor                    156.31 ns/iter   6,397,537.3 (148.54 ns … 211.13 ns) 157.49 ns 199.82 ns 208.23 ns
TransformStream constructor                   471.29 ns/iter   2,121,815.3 (452.53 ns … 791.41 ns) 468.62 ns 540.36 ns 791.41 ns
ReadableStream - iterator (2 chunks)            7.32 µs/iter     136,705.4   (6.35 µs … 639.97 µs)    7.1 µs  12.12 µs  20.98 µs
ReadableStream - iterator (20 chunks)          17.48 µs/iter      57,195.1  (14.48 µs … 289.06 µs)  16.06 µs  76.98 µs 114.61 µs
ReadableStream - reader (2 chunks)              6.86 µs/iter     145,847.9      (6.8 µs … 6.97 µs)   6.88 µs   6.97 µs   6.97 µs
ReadableStream - reader (20 chunks)            16.88 µs/iter      59,227.7  (14.04 µs … 311.29 µs)  15.39 µs  74.95 µs  97.45 µs
ReadableByteStream - iterator (2 chunks)        7.94 µs/iter     125,881.2   (6.86 µs … 811.16 µs)   7.69 µs  11.43 µs   16.6 µs
ReadableByteStream - iterator (20 chunks)      22.23 µs/iter      44,978.2  (18.98 µs … 590.11 µs)  20.73 µs  45.13 µs  159.8 µs
ReadableByteStream - reader (2 chunks)           7.4 µs/iter     135,206.9     (7.36 µs … 7.42 µs)    7.4 µs   7.42 µs   7.42 µs
ReadableByteStream - reader (20 chunks)        21.03 µs/iter      47,555.6  (17.75 µs … 357.66 µs)  19.52 µs  98.69 µs  146.5 µs
```

---------

Co-authored-by: Luca Casonato <hello@lcas.dev>

cli/tests/unit/streams_test.ts

@@ -126,9 +126,7 @@ function emptyChunkStream() {
   });
 }
 
-// Try to blow up any recursive reads. Note that because of the use of Array.shift in
-// ReadableStream, this might not actually be able to complete with larger values of
-// length.
+// Try to blow up any recursive reads.
 function veryLongTinyPacketStream(length: number) {
   return new ReadableStream({
     start(controller) {
@@ -249,9 +247,9 @@ Deno.test(async function readableStreamLongAsyncReadAll() {
 });
 
 Deno.test(async function readableStreamVeryLongReadAll() {
-  const rid = resourceForReadableStream(veryLongTinyPacketStream(10000));
+  const rid = resourceForReadableStream(veryLongTinyPacketStream(1_000_000));
   const buffer = await core.ops.op_read_all(rid);
-  assertEquals(buffer.length, 10000);
+  assertEquals(buffer.length, 1_000_000);
   core.ops.op_close(rid);
 });