feat: auto config cache size according to memory size

[QuenKar]

I hereby agree to the terms of the GreptimeDB CLA

What's changed and what's your intention?

• according to host memory size to decide the mito's buffer or cache size.
• remove cpu_nums crate and use sysinfo.

For example, my computer has 16G memory and 8 cpu cores

Mito(
    MitoConfig {
        num_workers: 4,
        worker_channel_size: 128,
        worker_request_batch_size: 64,
        manifest_checkpoint_distance: 10,
        compress_manifest: false,
        max_background_jobs: 4,
        auto_flush_interval: 1800s,
        global_write_buffer_size: 256.0MiB,
        global_write_buffer_reject_size: 512.0MiB,
        sst_meta_cache_size: 64.0MiB,
        vector_cache_size: 128.0MiB,
        page_cache_size: 128.0MiB,
        enable_experimental_write_cache: false,
        experimental_write_cache_path: "",
        experimental_write_cache_size: 128.0MiB,
        sst_write_buffer_size: 8.0MiB,
        scan_parallelism: 2,
        parallel_scan_channel_size: 32,
        allow_stale_entries: false,
    ...
    ...
    },
),

The possible concern here is sysinfo couldn't get system info in some unpopular OS.

---

update 2024.1.17
docker container(ubuntu 24.04), host memory=28G, cpus=16, we limit cpus = 8, memory = 8G, test again.
the result:

MitoConfig {
    num_workers: 4,
    worker_channel_size: 128,
    worker_request_batch_size: 64,
    manifest_checkpoint_distance: 10,
    compress_manifest: false,
    max_background_jobs: 4,
    auto_flush_interval: 1800s,
    global_write_buffer_size: 128.0MiB,
    global_write_buffer_reject_size: 256.0MiB,
    sst_meta_cache_size: 32.0MiB,
    vector_cache_size: 64.0MiB,
    page_cache_size: 64.0MiB,
    enable_experimental_write_cache: false,
    experimental_write_cache_path: "",
    experimental_write_cache_size: 64.0MiB,
    sst_write_buffer_size: 8.0MiB,
    scan_parallelism: 2,
    parallel_scan_channel_size: 32,
    allow_stale_entries: false,
    ...
    ...
},

Checklist

• I have written the necessary rustdoc comments.
• I have added the necessary unit tests and integration tests.
• This PR does not require documentation updates.

Refer to a related PR or issue link (optional)

[codecov]

Codecov Report

Attention: 2 lines in your changes are missing coverage. Please review.

Comparison is base (c2edaff) 85.63% compared to head (24603b8) 85.31%.
Report is 7 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3165      +/-   ##
==========================================
- Coverage   85.63%   85.31%   -0.32%     
==========================================
  Files         829      829              
  Lines      135773   136163     +390     
==========================================
- Hits       116273   116172     -101     
- Misses      19500    19991     +491     

[QuenKar]

@zhongzc I add the check about cgroup, if you have time, PTAL.

[fengjiachun]

I have an idea:
Can we obtain a logical memory size here, which should be smaller than the system memory.
Consider the following scenario:
The user deploys our DB on a machine with a large amount of memory, and this machine is also running other processes.
They want our DB to occupy less memory. It would be better if we can provide a parameter to specify the memory size, for example, by specifying it through startup parameters.
Then in the get_sys_total_memory method, we return the minimum value between it and sys_total_memory: Min(logical_total_memory, sys_total_memory).

The same idea for the get_cpus()