Repair lexicographic ordering among Bytes instances, which in turn repairs canonicalizing binary output, which in turn fixes the tests now there's just been a very long test name symbol added.

implementations/javascript/packages/core/src/bytes.ts

@@ -160,6 +160,11 @@ export class Bytes implements Preservable<never> {
         return nibbles.join('');
     }
 
+    valueOf(): string {
+        // Defined mostly in order to get lexicographic comparison to "work reliably"
+        return this.toHex();
+    }
+
     [PreserveOn](encoder: Encoder<never>) {
         encoder.state.emitbyte(Tag.ByteString);
         encoder.state.varint(this.length);

implementations/javascript/packages/core/test/reader.test.ts

@@ -26,6 +26,8 @@ describe('reading common test suite', () => {
             includeAnnotations: true,
             canonical: true,
         }));
+        // console.log('original', new Bytes(samples_bin).toHex());
+        // console.log('reencoded', new Bytes(bs).toHex());
         expect(bs).toEqual(new Uint8Array(samples_bin));
     });