Addressed boundary solution issues in UK employment alignment

[LiangShi369]

I have made following changes to SimPath Java code and Stata code:

For the simulation side:

Root causes of boundary roots after employment alignment:

1. Regression objects were built once and did not refresh when coefficient maps changed.
2. Single_Dep shares used BU‑level fracEmployed() (male+female / total), capping share at ~0.5.
3. Alignment for Single_DepMales/Females used fixed‑cost regressors or constant‑like terms, not affecting choice probabilities in fast path.

Key fixes:

1. Refresh regressions after coefficient updates
   • Parameters.refreshLabourSupplyUtilityRegression(...) added and called from ActivityAlignmentV2.adjustCoefficients(...).
2. Single_Dep share computed partner‑specific
   • In ActivityAlignmentV2.computeSimulatedShareUsingFraction(...), Single_DepMales uses male employment rate; Single_DepFemales uses female employment rate.
3. New alignment‑only regressors
   • AlignmentSingleDepMen, AlignmentSingleDepWomen added to BenefitUnit.Regressors.
   • Defined in BenefitUnit.getDoubleValue(...):
     • Men: couple, male at risk, female not, male employed.
     • Women: couple, female at risk, male not, female employed.
   • Subtracted in computeUtilityRegressionScoreWithoutFC(...), added back in updateFixedCostsAndLabour(...).
4. Single_Dep alignment switched to new regressors
   • SimPathsModel.activityAlignmentSingleDepMale/Female() uses AlignmentSingleDepMen/Women.
5. Subgroup-specific bounds
   • activityAlignment(...) now uses:
     • Couples: MAX_EMPLOYMENT_ALIGNMENT * 12
     • Single_DepMales: * 20
     • Single_DepFemales: * 20
     • Others: * 1
6. Comments added explaining rationale (bounds, refresh, single‑dep share logic).

For the data side:

1. 01_employment_shares_initpopdata do-file calculates the employment ratio of male/female in BUs that only one person is at risk of work:
   bu_fracemployed = bu_male_emp / (number of SingleDep_Males/Females BUs),
   Notice that unlike EU model, the UK model defines "at risk of work" additionally with "careNeedFlag!=1".

2. 02_person_risk_employment_stats do-file calculates employment statistics for all 7 categories (couple, single_male, single_female, singleDep_Males, singleDep_Females, singleAC_males and singleAC_females). It shows there are persons long-term disabled but employed, and persons need care but employed.