[ccomb]

Since #774 a score says what it rests on. Its contributions did not, yet they are quoted alone as often: "transport, 0.74 kg CO2 eq of the score", read from get_contributing_activities without score_activity first.

Now basis comes with:

• the contributing flows and activities of an impact category (REST and MCP);
• the contributing flows and activities of a single score (REST and MCP);
• the baseline of a sensitivity sweep: baseline.basis over REST, basis from compute_sensitivity.

What is left out, on purpose: aggregate and compare_activities give inventory amounts with no method, and the reduced model is a formula, not a figure.

• One way to read it: solutionBasis reads the basis off a solved score, the databases its chain reached. The score routes and get_impacts now use it too, instead of each spelling it out.
• The sensitivity tool scores the root's database alone (its own description says "root DB only"), so its basis names that database alone. The REST sweep goes through the dependencies, and its basis names them.

Wire revision 102. pyvolca decodes basis on ContributingFlows and ContributingActivities; SensitivityResult.baseline already did, being an LCIAResult.

Checks: cabal build all --ghc-options=-Werror, cabal test lca-tests, fourmolu, hlint (with the partials gate), and pytest on pyvolca.