✨ [RUM-10144] apply context defined just after init to events generated during init

[BenoitZugmeyer]

Motivation

Usually, context (global/user/account/feature flags etc.) is defined just after calling RUM.init(), which is an issue because events that are generated during init() will not have this context.

For now, this is only problematic for View events: RUM.init() generates a first View event that is missing the context defined right after. It is usually fine, as we can expect a RUM View Update to be generated at some point later. But sometimes no View Update happen for some reason (connectivity error, browser exiting abruptly...). In those cases, it is unexpected that the View isn't including the context.

This issue will be more pronounced when we'll start to collect errors that happen before init (in a future PR). In this case, we'll want the context to apply to those events (for example, to clearly identify which user was affected by such early error).

Changes

This PR fixes this issue by slightly delaying the events assembly. It uses a new data structure, BufferedObservable, that will be used to progressively replace the BoundedBuffer and what we call "pre start strategies". This BufferedObservable is asynchrone, so we leverage this fact to delay the assembly.

Test instructions

In the sandbox, define a user just after init():

      DD_RUM.init({
        clientToken: 'xxx',
        applicationId: 'xxx',
        ...
      })
      DD_RUM.setUser({ id: 'foo' })

Using the devtools extension, look at individual View events by unchecking Show only the latest View event in the "Events" tab.

When loading the sandbox URL, all View events should contain the @usr.id, even the very first View.

Checklist

• Tested locally
• Tested on staging
• Added unit tests for this change.
• Added e2e/integration tests for this change.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 92.38%. Comparing base (de2bc24) to head (3d012c1).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3592      +/-   ##
==========================================
+ Coverage   92.31%   92.38%   +0.06%     
==========================================
  Files         319      319              
  Lines        8104     8137      +33     
  Branches     1831     1834       +3     
==========================================
+ Hits         7481     7517      +36     
+ Misses        623      620       -3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[cit-pr-commenter]

Bundles Sizes Evolution

📦 Bundle Name	Base Size	Local Size	𝚫	𝚫%	Status
Rum	146.07 KiB	146.70 KiB	650 B	+0.43%	✅
Rum Recorder	18.02 KiB	18.02 KiB	0 B	0.00%	✅
Rum Profiler	4.63 KiB	4.63 KiB	0 B	0.00%	✅
Logs	50.26 KiB	50.33 KiB	74 B	+0.14%	✅
Flagging	0 B	932 B	932 B	N/A%	✅
Rum Slim	105.70 KiB	106.28 KiB	596 B	+0.55%	✅
Worker	23.59 KiB	23.59 KiB	0 B	0.00%	✅

🚀 CPU Performance

Action Name	Base Average Cpu Time (ms)	Local Average Cpu Time (ms)	𝚫
addglobalcontext	0.009	0.007	-0.002
addaction	0.032	0.022	-0.010
addtiming	0.006	0.005	-0.000
adderror	0.027	0.022	-0.005
startstopsessionreplayrecording	0.001	0.001	-0.001
startview	0.006	0.005	-0.001
logmessage	0.027	0.025	-0.002

🧠 Memory Performance

Action Name	Base Consumption Memory (bytes)	Local Consumption Memory (bytes)	𝚫 (bytes)
addglobalcontext	25.51 KiB	26.76 KiB	1.25 KiB
addaction	55.27 KiB	55.33 KiB	60 B
addtiming	26.36 KiB	25.44 KiB	-934 B
adderror	62.66 KiB	58.76 KiB	-3996 B
startstopsessionreplayrecording	25.13 KiB	25.10 KiB	-29 B
startview	429.15 KiB	431.25 KiB	2.10 KiB
logmessage	56.02 KiB	60.86 KiB	4.84 KiB

🔗 RealWorld

[BenoitZugmeyer]

Superseded by #3597