memtoc 0.1.1

CRAN resubmission

Replaced commented-out help examples with small, executable examples and regenerated the documentation. Logging examples now run in ordinary checks.
Restored temporary checkpoints so background polling returns its samples; nested monitors use separate files, removed by toc_mem() or mem_clear().
Include start/end observations in the reported peak when polling is enabled.
Added regression tests for nested sampling, cleanup, and peak bounds.
Removed the inaccurate Windows system requirement and clarified sampling and temporary-file recovery limitations.

The development labels below predate the initial CRAN submission (0.1.0).

Fixed background monitor sample retrieval on Windows. The monitor now saves checkpoints on every sample and reads from the persistence file when the background process is terminated.
Fixed test script using unexported format_bytes() function.

Parallel worker monitoring: New workers parameter in tic_mem() allows monitoring memory across parallel workers from the future package ecosystem.
- workers = "auto": Automatically detect future workers and child processes
- workers = "none": Only monitor the main R process
- workers = "children": Monitor main process and all child processes
- Integer vector: Explicit list of PIDs to monitor
Per-worker statistics: Results now include worker_stats data frame with peak memory usage for each monitored process.
New function mem_parallel_info(): Diagnostic function to check parallel backend status and detected workers before running a job.
Results now include n_workers field showing number of worker processes monitored.

Background polling: New interval parameter in tic_mem() enables continuous memory sampling via a background R process. This captures true peak memory even for short-lived allocations.
Memory trajectory: Results now include a trajectory data frame with timestamped memory samples when background polling is enabled.
System memory context: Trajectory includes system-wide RAM usage (sys_total, sys_avail, sys_percent). Warnings are displayed when system RAM exceeds 80% or 95%.
CPU time tracking: Results include cpu_user and cpu_system fields showing CPU time consumed during the tracked interval.
Crash recovery: New mem_recover() function retrieves memory samples from a crashed R session using persisted checkpoint files.
Capability detection: New mem_capabilities() function checks which features are available on the current system.
Diagnostics: New mem_diagnose() function provides detailed troubleshooting when background polling is unavailable.