Fix the sample progress and analyses count lagging by one analysis

[ramonski]

Description of the issue/feature this PR addresses

A sample counts its analyses from the catalog: getAnalysesNum reads the review state of each analysis from its brain, and getProgress is computed from that count. Both are catalog metadata on the sample, and the samples listing draws the progress bar from them.

An analysis promotes its submit and its verify to the sample from within its own after-transition event, which runs before the machinery that reindexes the analysis. The sample is therefore reindexed while the brain of the analysis that triggered it still carries the state it had before, and it stores a count and a progress that are one analysis behind.

The error is always the last analysis of a sample, which is the one that moves the sample, so it shows on every sample a laboratory ever submits or verifies.

Current behavior before PR

A verified sample with one analysis, read back from the catalog:

review_state:    verified
getAnalysesNum:  [0, 1, 0, 1]   # [verified, total, not_submitted, to_be_verified]
getProgress:     33

The analysis itself is verified and the sample is verified, but the sample's own metadata says nothing is verified and one analysis is still waiting. The progress bar reads 33% where the formula gives 66%. With two analyses it reads 60% instead of 80%.

Desired behavior after PR is merged

review_state:    verified
getAnalysesNum:  [1, 1, 0, 0]
getProgress:     66

And for two analyses, 80% verified and 40% submitted.

Why reindexing the state alone is enough

review_state is both a FieldIndex and a metadata column on every senaite catalog (BaseCatalog.INDEXES and COLUMNS), and the count is read from the metadata. reindexObject(idxs=["review_state"]) passes update_metadata=1 along with the index list, IndexQueue carries it through optimize(), and CatalogMultiplexProcessor.reindex hands it to _reindexObject. The brain the sample then reads is current, and nothing else about the analysis is reindexed twice.

The placement matters more than the scope. The queue merges the entries for one object at the position of that object's first entry, so reindexing the analysis after the sample has already been told about the transition is too late. It goes before anything is promoted.

Verification

• A single-analysis sample taken through receive, submit and verify reports [1, 1, 0, 0] and 66%; a two-analysis sample reports [2, 2, 0, 0] and 80% verified, [0, 2, 0, 2] and 40% submitted. Before the change, 33%, 60% and 40%. Reproduced across several fresh instances.
• bin/test-senaite -s senaite.core -t "Workflow*": 33 tests, 0 failures.
• bin/test-senaite -s senaite.core -t "Analysis*": 42 tests, 0 failures.