Improve reference index and clarify w_max terminology in docs

[gustavdelius]

Documentation-only improvements to make the website more useful for fisheries-science researchers and to make the maximum-size terminology consistent across the package. No behaviour changes.

Reference index (pkgdown/_pkgdown.yml)

• New "Overview: the mizer workflow" section at the top of the reference index. It frames the whole page as a five-stage pipeline — create → calibrate → tune dynamics → project → analyse — with links to the key function in each stage, so a reader arriving cold can see how the sections below fit together. (Anchored on the mizer-package topic, which moved out of "Classes".)
• Disambiguated related function families in the section descriptions: calibrate...() vs match...(), ...Biomass vs ...Number, and the plot...() variants (...2 = overlay two objects, ...Relative = ratio, ...ObservedVsModel).
• Fisheries-notation map in the "Calculating rates" section: getPredMort() ↔ M2, getExtMort() ↔ residual M, getFMort() ↔ F, getMort() ↔ Z, with an explicit warning that getResourceMort() is mortality on the resource, not a component of fish M.

w_max message wording (R/setReproduction.R, R/manipulate_species.R)

Error messages that referred to w_max as a species' "maximum size" now describe it as the upper size-grid boundary, consistent with w_max being a purely computational parameter (the biological maximum size is w_inf; the reproduction cut-off is w_repro_max). The three affected test assertions were updated to match.

Model-setup wrappers (R/newSingleSpeciesParams.R, R/wrapper_functions.R)

newSingleSpeciesParams(), newTraitParams() and newCommunityParams() now document why they set w_max = w_repro_max: because they do not yet set up stochastic growth by diffusion, no individual grows beyond w_repro_max, so no grid headroom above it is needed. This is deferred, not changed — to be revisited when the constructors gain a diffusion parameter (#339). Man pages regenerated.

Notes

• No behaviour changes; changes are limited to roxygen, message strings, _pkgdown.yml, and the matching test assertions.
• The affected tests (test-setReproduction.R, test-manipulate_species.R) pass, and all three constructors build with the documented w_max == w_repro_max invariant holding.
• The rendered docs/ site is intentionally excluded from this PR (following the repo convention of source-only commits); rebuild the site on merge as usual.

🤖 Generated with Claude Code