fix(serializers): correct byte length calculation for typed arrays

[Skn0tt]

Fixes PWTEST_MODE=service npm run ctest -- page-evaluate:97.

[mxschmitt]

For future reference:

This change corrects the byte length calculation for typed arrays. In typed arrays, the length parameter represents the number of elements, not the total number of bytes. Different typed arrays have different bytes per element (e.g., Int8Array uses 1 byte, Int16Array uses 2 bytes, Int32Array uses 4 bytes).
By dividing value.ta.b.length by ctor.BYTES_PER_ELEMENT, the code now correctly creates the typed array with the right number of elements, rather than potentially creating an array with an incorrect number of elements based on total byte length.
This is a subtle but important fix to ensure that typed arrays are reconstructed correctly during serialization and deserialization.

[github-actions]

Test results for "tests 1"

3 flaky

⚠️ [firefox-page] › tests/page/page-evaluate.spec.ts:424:3 › should throw for too deep reference chain @firefox-ubuntu-22.04-node18
⚠️ [chromium-library] › tests/library/inspector/cli-codegen-1.spec.ts:23:7 › cli codegen › should click @ubuntu-22.04-chromium-tip-of-tree
⚠️ [webkit-page] › tests/page/page-screenshot.spec.ts:345:5 › page screenshot › should work while navigating @webkit-ubuntu-22.04-node18

38888 passed, 805 skipped
✔️✔️✔️

Merge workflow run.