fix(instrumentation): fix web instrumentation span clock drifts

[legendecas]

Which problem is this PR solving?

Fixes parts of #3355

Short description of the changes

• This PR addresses the call sites in the fetch/xhr instrumentations that spans are created and end with hard-coded performance clocks.
• This PR should be complementary to fix(tracing): make spans resilient to performance clock drift #3384 / fix(tracing): make spans resilient to performance clock drift #3434, which provide a solution in the trace-sdk.

Type of change

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

• TODO

Checklist:

• Followed the style guidelines of this project
• Unit tests have been added
• Documentation has been updated

[codecov]

Codecov Report

Merging #3518 (603253c) into main (eecb800) will increase coverage by 0.03%.
The diff coverage is 98.63%.

❗ Current head 603253c differs from pull request most recent head 5b49af7. Consider uploading reports for the commit 5b49af7 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3518      +/-   ##
==========================================
+ Coverage   93.75%   93.79%   +0.03%     
==========================================
  Files         249      249              
  Lines        7610     7638      +28     
  Branches     1587     1588       +1     
==========================================
+ Hits         7135     7164      +29     
+ Misses        475      474       -1     

Impacted Files	Coverage Δ
...s/opentelemetry-instrumentation-fetch/src/fetch.ts	96.59% <95.83%> (-0.42%)	⬇️
...emetry-instrumentation-xml-http-request/src/xhr.ts	97.68% <100.00%> (+0.10%)	⬆️
packages/opentelemetry-core/src/common/time.ts	98.57% <100.00%> (+3.11%)	⬆️
packages/opentelemetry-sdk-trace-web/src/utils.ts	95.20% <100.00%> (+0.17%)	⬆️

[legendecas]

Closing in favor of #3434.