[refactor] Enable adaptive block_dim selection for CPU backend

[qiao-bo]

Related issue = #3750 #4541

For kernels with arch=ti.cpu, users tend to rely on the default block_dim settings. The default value is always 32 for CPU backends, which leads to slow execution for simple kernels. This PR adds the option to adaptive select block_dim during codegen. This leads to a 1.4x speedup for a simple kernel as mentioned in #4541:

@ti.kernel
def reduce_para()->ti.f32:
    n = v1.shape[0]
    sum = 0.0
    ti.loop_config(block_dim_adaptive=False)
    for i in range(n):
        sum += v1[i]*v2[i]
    return sum

[netlify]

✅ Deploy Preview for docsite-preview ready!

Name	Link
🔨 Latest commit	6b92a52
🔍 Latest deploy log	https://app.netlify.com/sites/docsite-preview/deploys/62ab3761c7bc480009b62361
😎 Deploy Preview	https://deploy-preview-5190--docsite-preview.netlify.app
📱 Preview on mobile	Toggle QR Code... Use your smartphone camera to open QR code link.

---

To edit notification comments on pull requests, go to your Netlify site settings.

[bobcao3]

LGTM!