Improve --centre_name and --upload_tpa arguments usage

Difficulty: Medium

Why this matters: --centre_name is currently required for every run of the pipeline against ENA (European Nucleotide Archive), in every mode. But according to ENA, it's only actually needed for "broker" accounts, accounts submitting data on behalf of somebody else. For a normal account, ENA works out the centre name automatically. On top of that, --centre_name is only actually used by the MAGs/bins workflow today, the metagenomic assemblies workflow ignores it completely even though the pipeline still asks every user for it. --upload_tpa has a similar scoping issue, it only matters for assemblies and MAGs/bins, not for reads.

Some terms: ENA (European Nucleotide Archive) is the public repository this pipeline submits data to, and Webin is ENA's account and authentication system needed to submit anything (free, self-service registration). A broker account is a Webin account that submits data on behalf of other organisations (for example a national data centre submitting on behalf of many research groups). TPA (Third Party Assembly) means an assembly built from data that was already public elsewhere, rather than newly generated, this is already explained well in docs/usage.md.

Skills / experience: mostly tracing where a couple of parameters are used across the codebase and cleaning that up, a good task if you want to get familiar with how parameters flow from the schema into the workflows. You'll also touch some tests, see the nf-core nf-test testing guide if nf-test is new to you.

Suggested approach:

1. Read the relevant parameter tables in docs/usage.md (mags/bins and metagenomic_assemblies sections) to see how --centre_name and --upload_tpa are currently documented, and the "Third party assemblies (TPA)" section for background on TPA.
2. Find every place --centre_name and --upload_tpa are referenced: nextflow_schema.json, workflows/assemblysubmit.nf, workflows/genomesubmit.nf, modules/local/genome_upload/main.nf.
3. Confirm the assemblies workflow really doesn't use centre_name (worth double checking, code may have changed since this was first noticed).
4. In nextflow_schema.json, make centre_name optional instead of required, and rename it (and the CLI flag) to --center_name everywhere it appears, including the README, docs/usage.md, and test configs.
5. Make sure the genome upload step behaves correctly when centre_name is left empty, it probably needs to just omit the argument rather than pass an empty string.
6. Double check --upload_tpa really has no use in reads mode and keep it scoped to the modes that need it.
7. Update tests and docs to match.