Ensure foreground stats are recorded for handled errors

[fractalwrench]

Goal

Ensures that foreground stats are recorded correctly for handled errors. This fixes a bug where app.inForeground, app.durationInForeground, and app.duration were captured for unhandled errors but not for handled errors. This is due to the two errors using a different mechanism for capture.

Changeset

• Added [BSG_KSCrash captureApplicationStatsInfo] to BSG_KSSystemInfo to ensure that application stats are always captured for handled errors
• Returned the value of bsg_kscrw_i_writeAppStats as a JSON string when capturing application stats
• Renamed BSG_ThreadDataBuffer to BSG_JSONDataBuffer as the struct can be used as a buffer for both threads + app stats

Tests

Verified locally in an example app that the fields are now populated.

[tomlongridge]

A few suggestions for consistency in naming, in case we use the data for other things in future.

It seems a little convoluted to encode and decode the JSON when all we're doing is reaching inside the crash context. It also calls this for any report, not just handled.

Did you consider using it directly with bsg_kscrashstate_currentState from BugsnagClient? Would it be better to override the AppWithState fields with this once it's been created for new handled errors only?

[fractalwrench]

Updated approach to avoid serialization by getting the information from the KSCrash state directly. This necessitates a small refactor of the logic for updating the duration/durationInForeground fields, which need to be recalculated whenever an error is captured.

[tomlongridge]

Looks good and works as expected now.

Small quibble over one of the method names and an extraneous declaration.