[Lang] Support linear system solving on GPU with cuSolver

[FantasyVR]

Related issue = #2906

We implement a basic linear system solver according to cuSolverSp_LinearSolver.

The demo is shown in misc/test_coo_cusolver.py. To solve a linear system, you could follow the steps as below:

1. Create three ti.ndarrays according to COO format, the datatype of row/column indices ndarray should be ti.i32 and the datatype of value ndarray should be ti.f32.
2. Create a ti.linagl.SparseMatrix as A.
3. Build the sparse matrix using build_coo function. The sparse matrix will be stored in CSR format.
4. Create a ti.linalg.SparseSolver.
5. Create two ti.ndarrray to store the right hand side b of linear system and the unkonw variables x.
6. Solve the linear system A * x = b with solve_cu function.

[netlify]

✅ Deploy Preview for docsite-preview ready!

Name	Link
🔨 Latest commit	777c5aa
🔍 Latest deploy log	https://app.netlify.com/sites/docsite-preview/deploys/630df34a965ad70009972e57
😎 Deploy Preview	https://deploy-preview-5860--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.

[Hanke98]

Sorry for the late response! Just flew back from a conference today. I will do a more careful review tomorrow!

[Hanke98]

Thanks a lot! LGTM mostly! I only have few nits:

[Hanke98]

LGTM! Thanks!