Add non-linear equality constraints in COBYLA

[emmt]

This PR fixes issues:

• Add non-linear equality constraints in COBYLA #14
• Make a link to https://github.com/robertfeldt/BlackBoxOptim.jl #16
• Scaling of variables #4

[emmt]

@zaikunzhang and @amontoison: Can you have a look at this PR before I merge it?

The musl Linux tests failed because of insufficient credits...

[amontoison]

@zaikunzhang and @amontoison: Can you have a look at this PR before I merge it?

I suggest to add a reference to the Julia package NOMAD.jl.

The musl Linux tests failed because of insufficient credits...

I checked the logs and we have some StackOverflowError.
For information, we have unlimited credits with CirrusCI if the repository is public @emmt.

[emmt]

I checked the logs and we have some StackOverflowError.
For information, we have unlimited credits with CirrusCI if the repository is public @emmt.

You are right, I read the log files too quickly. Sorry!

Yet I do not see why StackOverflowError only arises on the musl Linux platform. In Julia, this kind of error is usually the sign of an infinite function call recursion. I will investigate that.

[emmt]

I suggest to add a reference to the Julia package NOMAD.jl.

Good idea! Done in a73f854 of this PR.

[zaikunzhang]

Thank you @emmt and @amontoison . I give some comments at #14 for your consideration. In particular, #14 (comment)

BTW, due to some urgent deadlines, I will be almost away from GitHub until the end of the month. Thank you for your understanding.

[amontoison]

@emmt Should we remove Linux MUSL from CI?

[emmt]

@amontoison yes we should perhaps do that because Linux musl is not recommended (on Julia download page) for production use. The StackOverflow error my arise from issues not in PRIMA.jl code.

[github-advanced-security]

check-spelling found more than 10 potential problems in the proposed changes. Check the Files changed tab for more details.

[emmt]

Issue #4 is solved by this PR.

The only remaining issues are the check-spelling ones.

[emmt]

issuccess is a false positive, it is the name of this method imported from LinearAlgebra

[emmt]

All tests passed. I merge this PR and will trigger a new version (0.2.0) since the API has changed quite substancially...