Implementation/design improvements (v2)

[roll]

• Unify naming inferTableSchema vs inferSchemaFromTable?
• Rename saveSchema/Dialect to saveSchema/DialectDescriptor?
• Review normalize/denormalize in table
• Decompose core/general
• Move non-breaking to porting issue
• Rebase Promise.all on p-all with concurency
• Update dpkit link in readmes
• Use pl export for polars
• Replace pl.DataType by pl
• Rename df -> frame?
• Hide node warning in vitest
• Move session to cli/terminal root
• Rework dpkit/general/profile/scripts (use @ notation)
• Rename binaries to dpkit-terminal-… (rename dp command to dpkit command)
• Support musl on linux
• Rename dpkit validateDescriptor to validateMetadata?
• Increase dpkit tests to 80% coverage
• Move all the errors to dpkit/metadata?
• Add createReport function (errors, maxErrors)? -- will unify validate functions
• Consider Report and Report types for dpkit
• Improve dpkit’s error hierarchy (Error that includes everything in lib)
• Use inspect for function that just return errors?
• Split inspect/validateTable (low/high-level?)
• Split validate/inspectFile
• Rename library.dpkit export (clash with * possible export)?
• Simplify MetadataError?
• Add document.type == json support (with resource.jsonSchema validation)
• Add BoundError
• Rename profile to jsonSchema?
• Extract vitest-polly as a separate repo
• Merge file/dataset? (new name? dataset or data)
• Review when ts required rootDir (recursive imports?)
• Merge packages into bigger groups
• Rebase main on pre-release
• Use package.json/files to NPM publishing? (remove npm.ignore?)
• Create dataset/plugins/descriptor?
• Rename inferResourceFormat/etc? (shorter)
• Simplify plugin hierarchy (table/dataset)
• Consider moving fk validation back to table
• Fix compilations problems on the CI
• Fix table operations in compiled binaries
• Reduce amount of package exports in dpkit@2
• Increase test coverage to 90%