[experimental][hdl] Support for basic floating-point arithmetic units in the experimental backend

[Jiahui17]

This PR adds support for the basic floating-point arithmetic units in the experimental backend. This PR largely reused the changes made in #76, which uses flopoco-generated IP cores to replace the proprietary IP cores extracted from Vivado HLS.

Changes:

1. Flopoco-generated IPs are copied to the experimental RTL library.
2. addf, subf, mulf, divf, cmpf now use the flopoco-generated IP cores.
3. cmpf is now based on a more verbose opcode (cmpf-generator is also updated).
4. Dependency update in rtl-config.json
5. All the VHDL floating-point units now explicitly assert if operating on 32-bit data.
6. New benchmarks (integration-test) that use floating-point units.

Future work:

1. Characterization of flopoco-generated units
2. Flexible bit-width or number format

---

Bonus change:

1. CMake configuration flags are now added into .vscode/settings.json (those flags are identical to the ones in build.sh)

---

Known issues:

1. Some new benchmarks are not working with the experimental backend.

[Jiahui17]

LGTM! Thank you