fix: ensure all entries from bpf map are copied

[sthaha]

Previously only a MapSize (lesser than size of processes) of entries used to be copied from processes bpf map. The right use of a batch size less than the size of map is as follows:

Ref: https://lwn.net/Articles/797808/

    p_key = NULL;
    p_next_key = &key;
    while (true) {
       err = bpf_map_lookup_batch(fd, p_key, &p_next_key, keys, values,
                                  &batch_size, elem_flags, flags);
       if (err) ...
       if (p_next_key) break; // done
       if (!p_key) p_key = p_next_key;
    }

This PR fixes the issue by creating a entries byte array with the same size as the processes map.

Additionally, this commit makes use of bytes.Reader instead of bytes.Buffer to avoid unnecessary copy of ephemeral byte array.

[github-actions]

🤖 SeineSailor

Here's a concise summary of the pull request changes:

Summary: This PR fixes an issue where only a portion of entries were copied from the BPF map by creating an entries byte array with the same size as the processes map and using bytes.Reader instead of bytes.Buffer. The changes affect four functions: CollectCPUFreq, getCPUCores, resizeArrayEntries, and libbpfCollectProcessBatchSingleHash. The exporter struct is updated, and the libbpfCollectProcessBatchSingleHash function now uses processes.MaxEntries() instead of the hardcoded MapSize.

Key Modifications:

• Fixed incomplete entry copying from the BPF map
• Optimized code by avoiding unnecessary byte array copies
• Updated exporter struct and libbpfCollectProcessBatchSingleHash function

Impact: The changes do not alter the signatures of exported functions, global data structures, or variables, and do not affect the external interface or behavior of the code. The PR provides a targeted fix and optimization without introducing any breaking changes.

[sthaha]

go fmt artefact