Simplify expression and improve perf to get parent span id

[shaun-cox]

Changes

Avoids calling Context.has_active_span() followed by Context.span(), which is two lookups,
and instead relies on checking for an invalid span context on the return value of the latter.

Code is a little easier to understand and shows about 4-6% perf improvement in start-end-span benchmarks.

My assumption here is that a span with an invalid span context could never be made active in the first place.

Merge requirement checklist

• CONTRIBUTING guidelines followed
• Unit tests added/updated (if applicable)
• Appropriate CHANGELOG.md files updated for non-trivial, user-facing changes
• Changes in public API reviewed (if applicable)

[codecov]

Codecov Report

Patch coverage: 100.0% and project coverage change: -0.2 ⚠️

Comparison is base (0a893ce) 50.9% compared to head (52860aa) 50.7%.

❗ Current head 52860aa differs from pull request most recent head 0659f4c. Consider uploading reports for the commit 0659f4c to get more accurate results

Additional details and impacted files

@@           Coverage Diff           @@
##            main   #1088     +/-   ##
=======================================
- Coverage   50.9%   50.7%   -0.2%     
=======================================
  Files        165     165             
  Lines      19689   19789    +100     
=======================================
+ Hits       10025   10040     +15     
- Misses      9664    9749     +85     

Impacted Files	Coverage Δ
opentelemetry-sdk/src/trace/tracer.rs	93.4% <100.0%> (ø)

... and 21 files with indirect coverage changes

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[djc]

I think this is a good direction, but IMO it doesn't quite go far enough, so submitted #1089. Please have a look -- would be great to hear how that change affects your benchmarks.

[jtescher]

Hm I don't think this will have identical behavior to the previous implementation as children of invalid spans may be valid (have valid trace ids if not explicitly set to invalid in builder).

[jtescher]

Is the perf similar if you get the parent span context once and then check if it's valid subsequently, rather than checking if there is an active span before extracting the span context?

[hdost]

Superceded by #1089